Pressure Transients in Water Engineering

Pressure transients in
water engineering
A guide to analysis and interpretation
of behaviour
John Ellis
Independent Hydraulics Consultant and Honorary Research Fellow,
University of Glasgow, UK
Published by Thomas Telford Publishing, Thomas Telford Ltd, 1 Heron Quay, London E14 4JD.
www.thomastelford.com
Distributors for Thomas Telford books are

USA: ASCE Press, 1801 Alexander Bell Drive, Reston, VA 20191-4400, USA
Japan: Maruzen Co. Ltd, Book Department, 3—10 Nihonbashi 2-chome, Chuo-ku, Tokyo 103
Australia: DA Books and Journals, 648 Whitehorse Road, Mitcham 3132, Victoria
First published 2008
A catalogue record for this book is available from the British Library
ISBN: 978-0-7277-3592-8
# Thomas Telford Publishing Ltd 2008
Whilst every reasonable effort has been undertaken by the author and the publisher to
acknowledge copyright on material reproduced, if there has been an oversight the publishers
will endeavour to correct this upon a reprint.
The right of John Ellis to be identified as the author of this work has been asserted by him in
accordance with the Copyright, Designs and Patents Act 1988.
All rights, including translation, reserved. Except as permitted by the Copyright, Designs and
Patents Act 1988, no part of this publication may be reproduced, stored in a retrieval system or
transmitted in any form or by any means, electronic, mechanical, photocopying or otherwise,
without the prior written permission of the Publishing Director, Thomas Telford Publishing,
Thomas Telford Ltd, 1 Heron Quay, London E14 4JD.
This book is published on the understanding that the author is solely responsible for the statements
made and opinions expressed in it and that its publication does not necessarily imply that such
statements and/or opinions are or reflect the views or opinions of the publishers. While every
effort has been made to ensure that the statements made and the opinions expressed in this
publication provide a safe and accurate guide, no liability or responsibility can be accepted in
this respect by the author or publishers.
Typeset by Academic þ Technical, Bristol

Printed and bound in Great Britain by MPG Books, Bodmin, Cornwall
Contents
Acknowledgements xv
Notation xvi
Introduction 1
Chapter 1 Motivation for hydraulic transient analysis 7

1.1 Primary purpose of analysis, 7
1.2 Secondary objectives, 8
1.3 Permitted pressures, 8
1.4 Maximum pressures, 8
1.5 Pipe materials, 8
1.6 Rigid pipes, 9
1.6.1 Grey cast iron, 9
1.6.2 Asbestos cement, 10
1.6.3 Concrete pipes, 10
1.7 Flexible pipes, 11
1.7.1 Ductile iron, 11
1.7.2 Steel pipe, 11
1.8 Overpressure allowance, 12
1.9 Pipe linings for rigid and flexible pipes, 13
1.9.1 Bitumen, 13
1.9.2 Coal tar enamel, 13
1.9.3 Coal tar epoxy lining, 13
1.9.4 Cement mortar, 13
1.9.5 Paint systems, 14
1.9.6 Polyethylene lining, 14
1.10 Plastic pipes, 14
iii
1.10.1 Thermosetting plastics, 14
1.10.2 Thermoplastics, 15
1.11 Failure modes of pipes, 17
1.12 Maximum pressure and allowable amplitude of surge
in plastic pipes, 18
1.13 Minimum pressures, 18
1.14 When is analysis necessary?, 19
Chapter 2 Derivation of basic equations 21

2.1 The rigid-column approach, 21
2.2 Compressible flow theory, 23
2.2.1 Conservation of force, 23
2.2.2 Conservation of mass, 24
2.2.3 Compressible flow equations in terms of total head
H, 25
Chapter 3 Interpretation of a 27
3.1 Fluid properties, 27
3.2 Influence of the conduit wall, 28
3.3 Simple expression for a, 29
3.4 Variation of a with conduit shape, 32
3.5 Influence of gas on a, 32
3.6 The effect of sewage, 38
Chapter 4 Characteristic equations 41

4.1 Development of characteristic equations, 41
4.2 Significance of the integrals, 44
4.3 Effect of changing pipe elevation, 44
4.4 Pipeline resistance, 45
4.4.1 Corrosion, 47
4.4.2 Sliming, 47
4.4.3 Evaluation of the integral, 48
Chapter 5 Application of characteristic equations 49

5.1 Use of the characteristics, 49
5.2 ‘Natural’ characteristic mesh, 52
5.3 Using variable wave speed a, 54
5.4 Use of a larger time step, 55
5.5 Use of a fixed wave speed, 56
5.6 Distribution of free gas along the pipeline, 58
5.7 Model output, 59
iv
Chapter 6 Boundaries 60
6.1 Types of boundary, 61
6.2 Reservoirs and tanks, 62
6.3 Branches and changes in pipe properties, 64
6.3.1 Specific cases — number of pipes ¼ 1, 67
6.3.2 Specific cases — change of cross-sectional area, 67
6.4 Response of a large pipe or trunk main, 69
6.5 Actuated valves and pipeline fittings, 71
6.5.1 Terminal valves, 73
6.5.2 In-line valve, 74
6.5.3 Automatic control valves, 75
6.6 Use of more than one time step, 77
6.7 Non-reflecting boundary, 78
6.8 Other bifurcation conditions, 82
6.8.1 Bifurcation with operating valves, 82
6.8.2 Isolating valves, 83
6.9 Continuous drawoff, 84
Chapter 7 Valve closure in a simplified system 87

7.1 Instantaneous valve closure at t ¼ 0, 87
7.2 From 0 < t L=a, 89
7.3 L=a < t 2L=a, 90
7.4 2L=a < t 3L=a, 92
7.5 3L=a < t 4L=a, 93
Chapter 8 Actual pipelines 95

8.1 Attenuation, 95
8.1.1 Conditions at the wavefront, 97
8.1.2 Conditions when the wave height is of zero
amplitude, 99
8.1.3 Conditions at the closed valve, 100
8.1.4 Conditions downstream of a pump or valve, 100
8.2 A uniform gravity main, 100
Chapter 9 Valve operations 106

9.1 Treated water main, 107
9.2 Improving valve operation, 113
9.3 Two-stage valve closure, 113
9.4 Submerged discharge valve, 117
9.5 In-line valves, 118
9.5.1 Isolating valves, 118
v
9.5.2 Actuated valve, 119
9.6 Control of transient pressures and estimation of valve
operating time, 122
Chapter 10 Pumps 125

10.1 Types of pump, 125
10.1.1 Pumps which produce transient behaviour only when
changing their mode of operation — that is, starting,
stopping or changing speed, 125
10.1.2 Pumps which generate surge effects as a by-product of
their operation, 126
10.1.3 Pumps which produce transient events in order to fulfil
their function, 126
10.1.4 Pumps which do not by themselves produce surge
effects, 126
10.2 Turbine pumps, 126
10.2.1 Centrifugal or radial flow pumps, 127
10.2.2 Mixed or semi-axial flow pumps, 128
10.2.3 Axial flow or propeller pumps, 128
10.3 Turbine pump performance curves, 128
10.4 Including turbine pumps in hydraulic transient
analyses, 132
10.4.1 Transfer pump, 134
10.4.2 Booster pump, 136
10.4.3 Other pumping station and pipeline configurations, 137
10.4.4 Station losses, 139
10.5 System curves and pump duty, 139
10.6 Turbine pump start, 140
10.6.1 Direct start, 140
10.6.2 Star/Delta and transformer starting, 140
10.6.3 Variable speed or ‘soft’ start, 141
10.7 Case studies of pump start, 141
10.7.1 Simulation of direct start in solo pumping, 141
10.7.2 Direct start in multi-pump operation, 143
10.8 Initial conditions of flow, 146
10.9 Pump failure or ‘trip’, 146
10.10 Other pumps, 150
10.10.1 Reciprocating pumps, 150
10.10.2 Pneumatic ejector, 152
10.10.3 The hydraulic ram, 155
10.10.4 The jet pump, 156
vi
Chapter 11 Flywheels 159
11.1 Moment of inertia, 159
11.2 Flywheels, 160
11.3 Limitations on flywheel size, 161
11.4 Pipeline limitations, 162
11.5 Case study with different pump speed options, 163
11.6 Flywheels on a larger system, 167
11.7 Booster pump installations, 170
11.8 Multi-pump installations, 170
11.9 Advantages of flywheels, 171
Appendix Moment of inertia, 171
Chapter 12 Pressure vessels 173

12.1 Modelling a pressure vessel, 173
12.1.1 Polytropic relationship, 174
12.1.2 Rational heat transfer (RHT) equation, 176
12.2 Role of a pressure vessel in surge suppression, 176
12.3 Initial estimation of required pressure vessel volume, 177
12.3.1 Graphical techniques, 177
12.3.2 Simple numerical method, 178
12.3.3 More detailed numerical assessment, 178
12.3.4 Subsequent investigations and criteria, 178
12.4 Case study of a sewage pumping system, 179
12.5 Worst-case conditions, 181
12.6 Reversed flow and refilling a pressure vessel, 183
12.7 Low-lift systems, 189
12.8 Vessels at a booster pumping station, 193
12.8.1 The upstream pumping station, 194
12.8.2 The downstream pumping station, 196
12.9 Summary of response with a pressure vessel
included, 199
Appendix Equations for estimating air vessel parameters, 200
A12.1 Equation of motion, 201
A12.2 Solution ignoring resistance to flow, 203
A12.3 Including resistance to flow, 205
A12.4 Complete equations, 207
A12.5 Application of the equations, 207
A12.5.1 Maximum expanded gas volume, 207
A12.5.2 Peak upsurge pressure head, 209
A12.5.3 Required throttling, 212
A12.6 Pipeline system of varying cross-section, 214
vii
Chapter 13 Further aspects of pressure vessels 215
13.1 Pressure vessel types and their fittings, 215
13.2 Vessels having an air—water interface, 215
13.2.1 Air compressors, 215
13.2.2 Control of gas charge/liquid level, 217
13.2.3 Other vessel fittings, 218
13.3 Bladder vessels, 219
13.4 Positioning a pressure vessel, 221
13.5 Installation with air valves, 225
Chapter 14 Surge tanks and related structures 230

14.1 Purpose of a surge tank, 230
14.2 Simple analysis, 232
14.3 Long connection to a chamber, 233
14.4 Full-size connection, 236
14.5 Extent of protection, 236
14.6 Other aspects, 239
14.7 Initial estimates of surge tank parameters, 240
14.8 Related structures, 240
14.8.1 Service reservoir as a one-way surge tank, 241
14.8.2 Operation of an existing service reservoir, 242
14.8.3 Filtration plant, 243
14.8.4 Seawater intake system, 245
14.8.5 Seal weir, 248
14.8.6 Water towers, 249
14.8.7 Special structures, 253
Chapter 15 Feeder tanks or volumetric tanks 259

15.1 Components and location of a feeder tank, 259
15.2 Mode of operation, 261
15.3 Abnormal behaviour, 264
15.4 Mains duplication: Example 1, 267
15.5 Mains duplication: Example 2, 270
15.6 Aspects of feeder behaviour to consider, 276
15.7 Preliminary estimation of feeder tank volume, 277
Chapter 16 Discharge conditions 279

16.1 Vertical bellmouth, 279
16.2 A tank or chamber of finite area, 280
16.3 Back-flow connection, 282
16.4 Siphon breakers, 284
viii
16.4.1 Above-ground storage tanks, 284
16.4.2 Vacuum disconnecting valves, 286
16.5 Air valve operation, 292
16.6 Summary of influence of discharge arrangements, 293
Chapter 17 Air valves 295

17.1 Normal air valve locations, 297
17.2 Air valves for surge alleviation, 298
17.3 Events following flow reversal, 302
17.4 Air valve closure, 308
17.5 Case study of a sewage pumping station, 310
17.6 Pump restart with air in a pipeline, 314
17.7 Other considerations, 318
17.7.1 Uncertainties in simulation, 318
17.7.2 Liquid being conveyed, 319
17.7.3 Inspection, 320
17.7.4 Valve chamber and cover, 320
17.7.5 Valve materials, 321
17.8 Buffer tanks and estimation of required volumes, 321
Chapter 18 Air and gas 325

18.1 Pump start-up with an air-filled riser, 325
18.1.1 A more restricted air outflow device, 332
18.1.2 Soft-start of the pump, 333
18.1.3 Use of an accumulator, 334
18.2 Pump start with slow valve closure, 334
18.2.1 Air venting through a standard air valve, 336
18.2.2 A butterfly valve for air venting, 336
18.3 Air venting through a ‘sparg’ line, 339
18.4 Gas evolution, 339
18.5 Gas pockets in a pipeline, 340
18.6 Throttled outflow air valves, 343
18.7 Case study of a sewage rising main, 345
18.8 Pump blockage, 351
18.9 Pumped outfall pipeline, 354
18.9.1 Pipeline configuration, 354
18.9.2 Viking-Johnson coupling failure, 356
18.9.3 Hydrodynamic forces, 356
Chapter 19 Relief valves 359

19.1 Relief valve types, 359
ix
19.2 Initial valve sizing, 362
19.3 Valve positioning, 363
19.4 Analysis of behaviour, 363
19.5 Automatic surge control valve, 365
19.6 Surge anticipation valve, 366
19.7 Pumping station pressure relief valve, 366
19.8 Grove regulator, 369
19.9 High head relief valves, 371
19.10 Bursting disk, 375
Chapter 20 Check valve dynamics 376

20.1 Check valve response, 376
20.2 Pumping station check valves, 377
20.3 Consequences of an unsuitable check valve
installation, 377
20.4 Prediction of pumping station hydraulic
transients, 380
20.5 Reopening of a check valve door, 382
20.5.1 Check valve reopening due to pressure wave
reflections, 383
20.5.2 Valve reopening in longer term, 385
20.6 Check valve response in a multi-pump
installation, 388
20.7 Surge behaviour as a check valve shuts, 388
20.8 Modelling a pumping station, 390
20.8.1 Non-reflecting boundary with allowance for external
pipeline resistance, 390
20.8.2 System curve boundary, 393
20.9 Reduction of transient pressures following valve
closure, 394
20.10 Maximum pressures at a check valve, 396
20.10.1 Initial valve closure, 396
20.10.2 Cavitation upstream of the valve and resulting peak
pressures, 397
20.11 Other applications of check valves, 400
20.11.1 Check valve at the start of a rising main, 400
20.11.2 Check valve on vessel connection, 400
20.11.3 Bypass check valve, 400
20.11.4 Check valves along a rising main, 401
20.11.5 Inclusion of air valves with in-line check
valve, 406
x
20.11.6 Backflow check valve, 407
Chapter 21 Check valve characteristics 409

21.1 Check valve response, 409
21.2 Swing check valves, 411
21.2.1 Free-acting modifications, 414
21.2.2 Valve damping modifications, 415
21.3 ‘Recoil’ valves, 416
21.4 Tilting disk valve, 417
21.5 Rubber flap valve, 418
21.6 Split disk valve, 420
21.7 Butterfly valve used as a check valve, 421
21.8 Nozzle valves, 422
21.9 Moving ball, 424
21.10 Sleeve or duckbill valve, 425
21.11 Membrane valve, 434
21.12 Prediction of valve behaviour, 435
21.13 Use of the charts, 445
Chapter 22 Flexible pipe 446

22.1 Review of pipe materials and properties, 446
22.2 Pressure transient effects, 448
22.3 Strain and deflection, 449
22.4 Establishing the rate of ovalisation in the longer
term, 451
22.5 Long-term buckling pressures — unconstrained
surroundings, 452
22.6 Long-term buckling pressures — constrained
pipelines, 454
22.7 Deformation of a circular section and its effect on wave
speed, 458
22.8 Short-term elastic buckling under hydraulic transient
effects, 463
22.8.1 Unconstrained conditions, 463
22.8.2 Constrained conditions, 463
22.9 Application of a flexible pipe model, 465
22.9.1 Long horizontal pipeline, 465
22.9.2 Descending outfall, 467
22.9.3 Uniformly rising main, 468
22.9.4 Pipeline of differing properties, 470
22.10 Cyclical oscillations, 472
xi
Chapter 23 Amplification of transient pressures 474
23.1 Transmission of pressure waves through a branch
connection, 474
23.2 Pressure wave transmission through a change of cross-
section, 476
23.3 Meeting of opposing pressure waves, 478
23.4 Pressure waves in a suction main, 479
23.4.1 Protection of the rising main, 479
23.4.2 Conditions in the gravity main, 479
23.5 Amplification within a network, 481
23.5.1 Kirkleatham Lane Pumping Station, 481
23.5.2 System modelling, 484
23.5.3 Recorded pipe bursts and pressure extremes, 485
23.6 Wellfield transients, 488
23.6.1 Collector pipeline system, 488
23.6.2 Borehole and wellhouse configuration, 490
23.6.3 Wellfield operating conditions, 490
23.6.4 Pumpset inertia, 492
23.6.5 Sequenced pump operation, 493
23.6.6 Pumping failure, 494
23.6.7 Air valve operation, 496
23.7 Potential for amplification, 497
Chapter 24 Flow instabilities 499

24.1 Types of oscillation, 499
24.2 Pumping system — Glasgow East Main and Daer
network link, 501
24.2.1 Burnside booster pumping station, 503
24.2.2 Hydraulic transient computations, 503
24.2.3 Castlemilk Low pumping station, 503
24.2.4 Transient pressures, 504
24.2.5 Spread of unstable oscillations and consequences, 506
24.2.6 Possible remedies, 506
24.3 Gravity flow system, 507
24.3.1 The break pressure chamber, 508
24.3.2 Head losses, 508
24.3.3 Flow regulation, 508
24.3.4 H1 valve movement, 511
24.3.5 Remedial measures, 512
24.4 Small hydro station, 513
24.4.1 Observed behaviour, 513
xii
24.4.2 Comments on observed behaviour, 515
24.4.3 Modelling behaviour, 516
24.4.4 Remedial measures, 518
24.4.5 Final comments, 519
References 520
Further reading 525
Index 527
xiii
Acknowledgements
The author would like to express his appreciation for assistance received
from a number of sources. To Chris Bone of Morton & Bone Services,
UK agents for Charlatte bladder pressure vessels, who provided much
useful material on this form of surge protection. To Sam Gilbert of
Glenfield Valves Limited, Kilmarnock, Scotland, for many interesting
conversations regarding valve performance; also for providing infor-
mation on a wide variety of valve types. This book would not have
been possible without the work of engineers over many years who
laboured to improve understanding of this subject. From among now
retired engineers I would like to thank George A. Milne, formerly of
Crouch & Hogg, Glasgow, for taking time to discuss the subject at
length.
xv
Notation
A cross-section area of pipe measured normal to axis (m2 )

As horizontal water surface area
a speed of pressure wave through composite liquid/pipe
medium (m/s)
aH , bH , cH coefficients of equation Hp ¼ aH V 2 þ bH V þ cH
aT , bT , cT coefficients of equation T ¼ aT V2 þ bT V þ cT
C number of cycles till failure
Cp specific heat of gas at constant pressure
Cv specific heat of gas at constant volume
Cþ characteristic with gradient dx=dt ¼ V þ a (m/s)
C characteristic with gradient dx=dt ¼ V a (m/s)
c pipe circumference (m)
also pipe strength reduction factor
ca , cg pipe strength reduction factors
c1 pipe contraction restraint factor
c2 coefficient
D pipeline diameter (m or mm)
Df deformation factor
d handwheel diameter
dQ=dt rate of heat outflow from a pressure vessel
dr incremental change in radius
d incremental angle
E elastic (Young’s) modulus of pipe wall material (N/m2 )
E0 tangent modulus of soil
Ek kinetic energy of rotating body ¼ 12 I!2
Er elastic modulus of rock (N/m2 )
2
e P ¼ V =ð2gÞ
kinetic energy of flow in a pipeline
F overall head loss coefficient ¼ KL þ fL=D
also flow factor for regulator
xvi
Fro ¼ FVo2 =z
f friction factor in equation hf ¼ fLV 2 =ð2gDÞ
fc ovality correction factor P
fe equivalent overall friction factor ¼ KL D=L þ f
fos factor of safety
fs pipe support factor
G shear modulus of pipe material (N/m2 )
g acceleration due to gravity (m/s2 )
Hb inertial head rise in a branch pipe
Hbs base outlet head at a PRV
Hi inertial head rise ¼ aV=g (m)
Hj inertial head rise at a junction of pipes
Ho initial head or piezometric level
Hp head developed by pump
Hr liquid level in reservoir or large tank
Hv constant outlet head at a PRV or head rise at a valve
Hx inertial head rise at a point ‘x’ along a pipeline
h pressure head ¼ p=ðgÞ at a point in the pipeline (m)
also absolute pressure head in vessel
habs absolute pressure head
hatm atmospheric pressure head
hf head loss due to pipeline flow resistance (m)
hvap vapour pressure head (gauge)
I moment of inertia of a pumpset (kg/m2 )
Is ¼ s3 =12
i pipe number
i.l. invert level of pipeline
Jþ invariant or quasi-invarient quantity propagating along
Cþ path (m/s)
J invariant or quasi-invarient quantity propagating along
C path (m/s)
Jo initial value of invariant or quasi-invarient
K bulk modulus of fluid (N/m2 )
Kg bulk modulus of gas (N/m2 )
Ks bulk modulus of matter in sewage (N/m2 )
ki þ1 if isolating valve open and 0 if valve closed
ko equivalent uniform sand grain roughness height of new
pipe (mm)
ks equivalent uniform sand grain roughness height (mm)
L pipe length measured along axis (m)
lnðxÞ natural logarithm (to base e) of x
xvii
log10 ðxÞ common logarithm (to base 10) of x
m number of waves in buckling calculation
mg mass of gas dissolved/unit volume of liquid
N rotation speed of a pump (rev/min)
n polytropic exponent of gas law
Pr power output of a pump
p gauge pressure ¼ gh at a point in the pipeline (N/m2 )
pabs absolute pressure (N/m2 )
patm atmospheric pressure (N/m2 )
pb critical buckling pressure
pex external pressure
pH percentage hydrogen ion concentration in water (%)
pk critical pressure
ps set pressure at which regulator starts to open
Q flow rate at a pipeline section at some time (m3 /s)
Qair air inflow rate at air valve
Qe flow rate from an ejector system
Qm mixture discharge rate from a jet pump
Qo initial steady flow
Qout outflow to a gravity sewer
Qp flow rate of pumped medium through a jet pump
Qr discharge through relief valve or regulator
q ratio Qp =Qm for a jet pump
also vertical load on pipe/metre axial length
qi discharge at a demand point
R reactive force between sliding surfaces
Rc collapse rate of flexible pipe ¼ minutes/% decrease in
diameter
Re Reynold’s number of flow at a section ¼ VD=
rm distance from pivot to centre of gravity of check valve
door
rp radius of pivot pin
S gradient of head loss due to pipeline resistance
So initial steady flow gradient of head loss due to resistance
Sr ring stiffness/unit length of pipe
s pipe wall thickness (m)
sf sliming factor
sj þ1 for Cþ characteristic and 1 for C characteristic
sv Vo =jVo j provided jVo j > 0
T torque
Tabs absolute temperature
xviii
Teff effective valve closure time
Tr resultant thrust on valve hinge
t time (s)
V mean velocity of flow at a cross-section at some time
(m/s)
Vo initial velocity
Vol volume of liquid þ volume of gas ¼ Voll þ Volg (m3 )
Volf volume of free or undissolved gas (m3 )
Volg volume of gas dissolved in liquid (m3 )
Volin inlet volume of sleeve valve
VolL volume of liquid (m3 )
Volp volume of a pipeline
Vols volume of matter in sewage (m3 )
Volt total volume (m3 )
Ws buoyant weight of check valve door
x axial distance along pipeline from some datum (m3 )
Z head relationship for jet pump
z pipe centreline level measured from a common
horizontal datum (m3 )
also d/spreservoir level pressure vessel level þ habs
z ¼ Vo = ½LA=ðgAs Þ
zw weir crest level
¼ y=Dm
k rate of increase in roughness height with time
x distance increment in numerical scheme
xp small distance increment
t time increment in numerical scheme
tp small time increment
y vertical change in pipe diameter
! rotation speed change over a time increment
relief valve diameter
void fraction ¼ ð1 þ f Þðpatm =pabs Þð1=Þ

¼ Dmin =Dm ¼ 1
f void fraction of free gas ¼ Volf =ðVolf þ Voll Þ
s sewage matter void fraction ¼ Vols =Volt
1 void fraction of free gas
2 void fraction of dissolved gas ¼ Volg =ðVolg þ Voll Þ
ratio of specific heats for gas ¼ Cp =Cv
h horizontal radial deformation of pipe
i initial pipe deflection
xix
y vertical radial deformation of pipe
" maximum strain
"cr critical strain for a perfectly round pipe
"f effective maximum strain
pump efficiency
angle between flats of a polygonal duct (rad)
also angle of pipeline with the horizontal
also angle of check valve door from the vertical
Poisson’s ration for pipe wall material
also coefficient of sliding friction
kinematic viscosity of fluid (m2 /s)
ratio hm =h
liquid density (kg/m3 )
d density of disk material
g density of gas (kg/m3 )
m mean density of mixture ¼ g þ ð1 Þ (kg/m3 )
also mean density of mixture exiting from a jet pump
p density of pumped medium through a jet pump
s density of sewage matter (kg/m3 )
circumferential hoop stress in pipe wall (N/m2 )

also axial spring force
! rotation speed of a pump (rad/s)
!o rotation speed at the start of a time increment
!s specific speed of a pump
xx
Introduction
A number of terms have been associated with the phenomenon of

transient motion within pipeline and tunnel networks and it may be
worthwhile to attempt a definition of some of these expressions. As
far as the writer can discover, there are no universally agreed definitions
for some of these terms.
Unsteady flow is a general term to describe any motion which
embodies a time-varying component.
Waterhammer has been used by Streeter and Wylie (1967) as synon-
ymous with unsteady flow but embodying the word ‘hammer’ suggests
noise and can be specifically associated with a fast-changing pressure
which imposes a force on pipework and valves sufficient to generate
sound. Joukovsky (1898) was one of the first to analyse ‘waterhammer’
waves in which flow varies spatially as well as over time. Jaeger (1977)
defined waterhammer as a rapid oscillation.
The expression transient flow was used by Streeter and Wylie (1967)
to indicate a change from one state of steady motion to another.
Steady oscillatory flow, periodic flow or pulsatile flow all define condi-
tions of flow which are repeated at intervals of time called the period
of oscillation.
Resonance describes motion which can progressively develop into a
steady oscillatory flow.
The term surge has been associated with the concept of mass oscilla-
tions in which although flow is changing over time is essentially constant
along a pipeline. This expression often has been used in connection
with longer-period transients in a low-pressure tunnel system for a
hydropower plant whereas waterhammer has been used in relation to
faster transients in the high-pressure parts of hydroelectric plant.
Jaeger (1977) defined mass surges or mass oscillations as slow transients.
1
Pressure transients in water engineering
The distinction between surge or slow transients and waterhammer

or fast transients arose partly from the different methods of analysis
which have been employed in the past. For instance, the analysis of a
long-period surge or mass oscillation in a low-pressure conduit would
usually be analysed using relatively large time steps to reduce the
effort required and the faster waterhammer phenomenon necessarily
studied with much smaller time steps.
In common with many other branches of engineering, hydraulic tran-
sient or pressure surge analysis has been transformed over the past 20
years or so by the increasing availability and power of digital computers.
This has led to a considerable blurring of distinction between slower and
faster transients as the same computer program can be used to investi-
gate pressure transient behaviour throughout a system. The title
Pressure Surge has been used for the long series of international confer-
ences organised by the British Hydromechanics Research Association
(BHRA) which includes both long-period and fast transients. In this
book the terms pressure surge and hydraulic transient are considered to
be interchangeable and applied to all aspects of behaviour.
Concurrent with hardware developments, increasingly sophisticated
computer modelling programs have been introduced which allow a wide
range of pipeline network configurations to be represented. Whereas in
the past a substantial factor of safety would be applied to any design,
research has improved understanding of transient behaviour allowing
reduction in the factor of safety. Better understanding both helps to
prevent damage and also permits cost savings through a reduction in
the factor of safety. Continued pressure to reduce costs makes use
of the properties of pipes and fittings to a greater extent and also
drives the computational ability to analyse systems in increasing
detail. In the past, limits on ability to analyse systems may have led
to the adoption of simpler pipeline arrangements whereas today compu-
tational ability and understanding of behaviour allows pressure surge
events to be studied in even very complicated networks.
User-friendly graphical interfaces have been included in many of
these programs which permit the user to select system components
from a range of elements. Thus the transient analysis of quite compli-
cated pressure pipeline systems has increasingly moved from being an
activity undertaken by a fairly limited number of specialist organisations
towards becoming a capability within design offices generally.
Within specialist consultancies a wealth of knowledge will have been
accumulated which is at the client’s disposal but in order that the client
should be able to appreciate detailed aspects of system behaviour it is
2
Introduction
desirable that he should have on his team someone familiar with aspects
of transient behaviour. The better informed the client’s representative
the less scope for misunderstanding.
However, until now this type of analysis remains an activity which is
not encountered on a very regular basis within consultancy organisa-
tions generally and a considerable time may elapse between one
analysis and the next. Coupled with turnover of staff, this can make
the process of gaining experience somewhat protracted. The new
graduate or novice engineer may be provided with a quite sophisticated
modelling tool but still be unsure of what design cases should be consid-
ered and for how long a computer simulation is required to be continued
to ensure that all transient activity of significance has been predicted.
Being new to surge analysis, the novice may be uncertain whether or
not the predictions which he has obtained from the modelling
exercise are realistic, or possibly contain the consequences of some
mistake in the information provided to the computer or even the
result of using the program inappropriately. These concerns are not
unique to pressure transient analysis and have also arisen in connection
with the use of computer software in other areas of engineering such as
structural analysis where similar concerns have been raised.
However, the frequency with which transient investigations are
undertaken may increase generally if a novel approach to mains con-
dition assessment using controlled surges becomes widespread. As
described by Misiunas et al. (2007), this technique is in the early
stages of development but offers hope of a low-cost means of mains
assessment.
In a recent article, Gordon (2004) described how a computer
program written by a mathematician without adequate understanding
of engineering had resulted in expensive repairs and several months
delay in commissioning of a scheme. This article highlighted the
dangers of subcontracting the hydraulic transient study based on
lowest cost, without determining whether the program was adequate
for the task and had a proven track record on other projects. The
client had not worked on a project of this kind for a few years and
experienced personnel were no longer available. In another instance,
Gordon (2006a) described how a high head turbine runner was fabri-
cated with an even number of blades with the corresponding number
of wicket gates also being even. The basic principle laid down back in
1941, that an even number of wicket gates should be accompanied
by an odd number of runner blades, had been lost. The result was
that a large number of blades were cracked after a couple of months
3
in service, leading to an expensive repair job. The failure to appreciate

this long-established principle was attributed to the retirement of nine
engineers of long experience, and design being then left exclusively
in the hands of younger personnel all under the age of 30. Gordon
challenged industry to come up with a handbook containing recent
examples of projects and equipment.
A considerable body of literature exists on the topic of pressure tran-
sient analysis including texts which derive the basic equations of motion
and show how these equations can be used to develop practical
methods of computation for implementation on the digital computer.
The manner in which various system components, such as valves,
pumps and turbines for example, can be incorporated into the model-
ling process has also been described.
Many engineers new to the area of hydraulic transient prediction will
nonetheless have been exposed, at least in part, to the theory and
concepts underlying the models which they will be required to use in
conducting analyses of pipeline and tunnel systems. In this volume it
is not the intention to duplicate the material contained within these
existing texts to any appreciable extent but rather to describe and
examine a range of circumstances encountered during many years of
practice in this field. This book should therefore be considered as
complementary to these other texts which introduced the basic
concepts and theories. Where equations are derived or introduced,
this has been done with the intention of adding to the description of
different aspects of pressure transient behaviour. Some programs have
the capacity to permit additional ‘modules’ to be added to facilitate
modelling of features which are not available in the original program.
One or two examples are given of less common features together
with relevant equations which may be used to develop extra modules
for inclusion in programs.
On occasion, hydraulic transient analysis of a pipeline system has
been left until the design and even construction process is well
advanced. Many factors such as pipeline route, pipe material and
diameter, and types of equipment such as valves may all have been
chosen on the basis of steady flow analyses and prior to any considera-
tion of transient analysis and cannot readily be altered without financial
penalty. Yet a transient investigation which demonstrates the need for
surge protection of a network can also be used to examine a variety of
different options which may include alternative pipeline routes, pipe
materials and diameter for instance, which may possibly show that
adoption of a particular pipeline configuration may reduce or even
4
Introduction
remove the requirement for surge protection. If understanding of

hydraulic transient behaviour generally were improved then an initial
transient analysis undertaken in parallel with study of steady flow
behaviour, may yield savings in overall costs or some improvement in
network behaviour through selection of more suitable equipment.
Where preliminary investigations of a pipeline system have demon-
strated that unacceptable pressure surge behaviour is likely to
develop, differing philosophies may exist regarding how to control
these surges. The philosophy which is adopted may depend upon
whether one was brought up in what may be termed the American or
European school (McCrone, undated). Once a surge has developed
in a pipeline, American practice relies upon extensive use of special
surge control valves to limit or minimise the effect of these transient
pressure conditions. In contrast, the European approach is more
orientated towards prevention rather than cure, with emphasis on
avoiding the spread of unacceptable surging in the first instance.
With the emergence of very large global engineering firms, perhaps
with employees from both traditions, it may be important to have
some understanding of these alternative approaches. Some considera-
tion has been given to application of surge control measures from
both sources.
This book is intended for engineers who are relatively new to the
topic of pressure surge analysis and has been written in the hope that
it will be helpful in avoiding some of the pitfalls that await. After consid-
ering the requirement for transient analysis and possible objectives of
such studies, consideration is then given to derivation of pertinent
equations. Two methods of analysis of hydraulic transient behaviour
are discussed. Equations derived from ‘elastic theory’ have been used
for purposes of illustration of transient phenomena throughout the
book and it has not been the intention to review other techniques of
analysis such as the impedance method or graphical approaches.
These alternative methods have been amply covered in existing texts.
The exception has been use of the ‘rigid column’ analogy which has
particular uses with regard to estimating preliminary dimensions for
pressure vessels and surge tanks.
Implementation of the ‘elastic equations’ is then considered with
discussion of alternative methods of numerical implementation on
the digital computer. ‘Boundary conditions’ are introduced covering
some of the more common features of pipeline systems.
Application of the ‘elastic’ approach to a simple pipeline, neglecting
the influence of pipeline resistance and gradient, is discussed. This is
5
followed by a study of actual simple pipelines to highlight some of the

differences in behaviour which can occur.
A series of chapters is included in which the relative merits of alter-
native forms of surge protection are discussed together with case studies
of their application. Included are chapters on the behaviour of air and
gas in pipelines and the attendant dangers if even modest pockets of gas
are allowed to develop.
Problems associated with the selection of an inappropriate check
valve are considered and some guidance is given on choosing a valve
to suit a particular situation.
The volume concludes with consideration of some aspects of systems
which can produce departures of predicted conditions from observation.
Included is damping of pressure waves in flexible pipes and also
amplification of surge pressures either as a consequence of the system
configuration or through development of a resonant condition. Case
studies based upon actual systems have been included to illustrate
some of these aspects of hydraulic transient behaviour. Some aspects
of hydraulic transient analysis have not been included in this volume
but are amply covered elsewhere. A case in point is modelling of
condensers of power plants, detailed discussion of which may be
found in the recent IAHR Synthesis Report (2000) which is the
result of two decades of study into the phenomenon of hydraulic tran-
sients.
Actual design flows have been used in the majority of examples
throughout the book but where it was considered that a particular
phenomenon could be more clearly demonstrated, discharge conditions
have been allowed to vary to some extent from actual parameters.
6
1
Motivation for hydraulic
transient analysis
In times past when methods of analysis and materials were perhaps less
clearly understood than now or quality control was less exacting, a
relatively large factor of safety would be applied to any design. With
increasingly sophisticated computer modelling capabilities available
and with new materials available and a better understanding of these,
there has been a movement towards more detailed analysis and better
use of materials. Strengths of pipes and pipeline fittings are of para-
mount importance in determining whether a design is safe and so this
chapter examines some aspects of pipe materials and allowable internal
pressures before proceeding to consider analysis of pressure transients in
pipeline systems. Also included is some mention of pipe linings as these
can have a bearing upon allowable pressures. Other aspects such as
trench conditions and flexibility which can influence allowable pres-
sures will be considered at a later stage.
1.1 Primary purpose of analysis

Pipelines have been described as one of the arteries of modern civiliza-
tion and so protection of these is of great importance. The primary
concern underlying the majority of studies into unsteady flow behaviour
in pipelines is to establish a safe operating environment. This usually
involves analyses designed to predict extremes of pressure, both
maximum and minimum, which may potentially develop within the
system during its operating lifetime and the probability of occurrence
of each extreme event. These extremes may be the consequence of
planned or accidental changes in flow. A transient event can produce
pressures which exceed those maxima developed during steady flow
and also pressures below the anticipated minimum steady flow
7
pressures. Steady flow is also taken to include static or no-flow

conditions.
1.2 Secondary objectives

Additional information which may be required of a hydraulic transient
investigation can include operating times for equipment such as pumps,
turbines and valves, rates of filling or emptying of tanks and vessels and/
or necessary volumes of these tanks. Other pertinent information
relates to the timescale of the transient event, such as the time taken
for a steady flow to become established following a pump start or the
time for quiescent conditions to develop after some change in flow,
say after an alteration of valve setting. The analysis will also yield
parameters for the selection of suitable equipment such as self-acting
valves, for instance air valves and check valves, in order to avoid
unnecessary or unacceptable transient behaviour.
1.3 Permitted pressures

When considering maximum and minimum pressures, these must be
compared with defined maximum pressures. These allowable pressures
are determined by the strength of the pipeline, by the nature of the
liquid being conveyed and possibly by the nature and characteristics of
any pipe lining. For example, a relatively brittle lining such as cement
mortar may crack if pipe deformation is too great. Any crack in the
lining will potentially allow pipe contents to attack the pipe wall itself.
1.4 Maximum pressures

Regarding allowable peak pressures these are usually defined by the
supplier in accordance with codes of practice. For instance, with
respect to pipes the allowable maximum sustained pressure or long-
term pressure can be found in manufacturers’ literature. Likewise for
fittings the permissible pressure is readily obtainable from suppliers.
1.5 Pipe materials

Pipes come in a range of materials each having particular strengths and
weaknesses, from cast iron, which has been in use for well in excess of a
century, to medium density polyethylene (MDPE) introduced little
more than 20 years ago. Analysis often has to be undertaken on pipeline
8
Motivation for hydraulic transient analysis
systems comprising a mix of old and new pipes and the properties of
both types are important as well as any deterioration in these over
time. Since pipeline systems will usually be in service for many years
and be subject to extension and upgrading, to suit changes in
demand for instance, an understanding of properties of older or even
historic materials must still be retained as some reanalysis is often
necessary to accommodate changes to a network.
1.6 Rigid pipes
1.6.1 Grey cast iron

One of the earliest materials was grey cast iron (CI), with the first
recorded use being in 1455. Some of the cast iron pipes installed at
the Palace of Versailles in 1664 were removed relatively recently and
found to be in reasonable condition. Other pipes at Versailles are still
in service. Cast iron contains 3.5% carbon in the form of graphite
flakes and the pipe is designed as a rigid structure under combined
loadings from earth pressure, traffic, impact and transient pressures.
Various types of pipe joints have been provided including spigot and
socket, flanged, rubber gasket and joints with locking devices to resist
longitudinal thrust. Unlined pipes are subject to tuberculation with
nodular granules frequently developed in corrosive waters. Continued
exposure results in a gradual reduction in carrying capacity. Cast iron
has been produced in several classes. Some details of ratings and test
pressures are given in Tables 1.1 and 1.2.
Table 1.1. Ratings and test pressures for various classes of cast iron
Nominal Description Maximum working Recommended

internal pressure rating maximum site test
diameter including surge pressure
(mm)
bar g mWG bar g mWG
80—500 Spigot and socket 10 100 16 160

centrifugally cast iron pipe
class 1
80—500 Spigot and socket 16 160 25 250
centrifugally cast iron pipe
class 3
80—200 Standard fittings 12.5 125 20 200
mWG ¼ metres water gauge
9
Table 1.2. Test pressures for various classes of cast iron
Nominal internal Description Works proof test pressure

diameter (mm)
bar g mWG
80—500 Spigot and socket centrifugally cast 35 350

iron pipe classes 1 and 3
80—200 Standard fittings 20 200
Cast iron fails catastrophically.
1.6.2 Asbestos cement

Asbestos cement (AC) as a pipe material was invented around 1930.
Pipes are usually manufactured in diameters up to 900 mm. Several
layers of asbestos fibres are soaked in cement. Being entirely non-
conducting it was thought impervious to electrolytic corrosion although
being cement-based AC is susceptible to attack by acidic, soft or
sulphate-bearing waters and soils. Asbestos cement is a brittle material
which can fail catastrophically. Pressure may be defined by pressure
classes B, C and D for pressure head ranges 61—122 mWG or alterna-
tively as shown in Table 1.3.
1.6.3 Concrete pipes

Concrete pressure pipes can be classed as rigid or semi-rigid depending
on design. Pipe design pressures will commonly be from 14.1 kgf/cm2 .
This includes a 7.6 kgf/cm2 transient pressure allowance. Large-
diameter prestressed pipes are produced by a few manufacturers and
are available in larger diameters of 700—1200 mm and for heads up to
180 mWG. These become relatively more expensive at pressures
>17.6 kg/cm2 . These pipes tend to be heavy and may come in shorter
Table 1.3. Pressure classes for asbestos cement
Description Class
15 20 25
Test pressure (bar g) 15 (153 mWG) 20 (204 mWG) 25 (255 mWG)

Maximum working 7.5 10.0 12.5
pressure (bar g)
10
lengths thus requiring to have a larger number of joints. Unlike other

pipe types, concrete pipes cannot be cut to length. Compression-type
rubber gaskets are a fairly standard form of joint.
1.7 Flexible pipes
1.7.1 Ductile iron

Ductile iron (DI) is designed as a flexible pipe and has largely replaced cast
iron. It contains 3.5% carbon in the form of spheroids or nodules and has
similar corrosion characteristics as cast iron. It is a ductile material with
strength characteristics more like those of steel. Large-diameter pipes
are designed to withstand external loads as well as internal pressures.
Minimum wall thickness assumes a factor of safety against collapse
under vacuum pressures and provides working pressure ratings of at
least 14.1 kgf/cm2 þ 7 kgf/cm2 transient pressure allowance. Pipes may
be provided in diameters ranging to 1600 mm or more. Typically allowable
pressure falls with increasing diameter as illustrated by Table 1.4.
For cement-lined pipes allowable deflection is largely governed by the
need to avoid cracking of this lining. Suppliers will be able to provide
data on permitted deflections. In the case of Stanton Integral DI
pipes for example, the ratio of allowable vertical deflection to
diameter D is expressed as a function of pipe size in Table 1.5.
Deflection can be obtained from the modified Spangler formula.
1.7.2 Steel pipe

Steel pipe is produced in various grades. Yield points are typically in the
range 1760—2950 kgf/cm2 with still higher strengths available. Pipe
design for normal waterworks operations may use a working tensile
stress 50% of yield point stress. Typical diameters may range from
Table 1.4. Allowable pressures for various pipe diameters
Nominal internal diameter (DN)
80—300 350—600 700—1200 1400—1600
Test pressure 50 40 32 25
(bar g)
Maximum working 40 25 25 25
pressure (bar g)
11
Table 1.5. /D as a function of pipe size
Integral pipe size Maximum % deflection =D
DN 100 1.6
DN 150 2.1
DN 200 2.4
DN 250 2.7
DN 300 3.0
DN 350 3.1
DN 400 3.2
DN 500 3.4
DN 600 3.6
DN 700 3.8
DN 800 and above 4.0
150—3600 mm with more common sizes being in the range 400—

1500 mm. At working pressures <10.5 kgf/cm2 , strength is often
controlled by external load, requiring a certain minimum thickness to
guard against excessive deflection and collapse under vacuum pressures.
Common practice in the past has been to limit wall thickness to a
minimum of 6 mm for diameters up to 600 mm. While providing an
additional corrosion allowance, this thickness will frequently withstand
>17.6 kgf/cm2 and for high-strength steel considerably higher pressures.
Advances in cathodic protection and coating technology has made
possible wall thicknesses <6 mm. For larger pipe diameters and pres-
sures >10.5 kgf/cm2 , tensile strength often controls wall thickness.
Steel pipe test pressures vary according to the grade of steel. For
grades 22 to 27, pressures range from 70 kgf/cm2 (700 mWG) for
smaller diameters to 20 kgf/cm2 for larger diameters.
The maximum allowable deflection of a steel pipe should be not more
than 2% where a cement mortar lining and/or coating has been
provided. Concrete coating provides negative buoyancy for underwater
installation.
Older steel pipes in cold climates can potentially fail catastrophically.
Stainless steel pipe is also available and has the advantage of not
requiring corrosion protection. It is virtually maintenance-free and
relatively vandalproof.
1.8 Overpressure allowance

Many types of pipe are permitted, by codes of practice, to carry an over-
pressure allowance for transient or short-term events. For instance,
12
British Standards allow pipe materials, such as asbestos cement (AC),

ductile iron (DI) and steel, to carry peak transient pressures as high
as 115% of the maximum sustained working pressure.
1.9 Pipe linings for rigid and flexible pipes

To protect pipes from corrosion a range of lining materials are available
having different advantages and disadvantages. Types of lining are:
bitumen, coal tar, coal tar epoxy, epoxy resin, cement mortar, paint
systems and polyethylene. The lining can influence the outcome of
pressure surge investigations in a number of respects. For example,
pipe lining determines to some extent the resistance to flow experi-
enced and therefore the head discharge relationship for the system.
Furthermore, the strength of the lining may dictate allowable pressures,
for example where a cement mortar lining is used.
1.9.1 Bitumen
Bitumen linings comprise a mixture of 80% bitumen and 20% dry lime.
The bitumen deteriorates with age. Burstall (1997) reported ‘puffing-
up’ of bitumen to give a rough tuberculated appearance. Deterioration
can also result in loss of lining in raw water pipelines.
1.9.2 Coal tar enamel

Coal tar enamel was used up until the 1960s but its use in potable water
mains was discontinued on health and water quality grounds. Because of
resistance in low pH waters it finds use in raw water lines serving wellfields.
1.9.3 Coal tar epoxy lining

Coal tar epoxy lining is lighter than other lining systems and may be
used in pipe crossings where weight is a consideration. It can leach
chemicals into potable water and so its use is not preferable.
1.9.4 Cement mortar

Cement mortar is a standard lining for ductile iron and steel pipes and
for relining of cast iron mains. In some parts of the world it is perhaps
the only lining system which has a proven track record over many
years and so may be preferred over alternatives for that reason. In
13
treated water mains a long life can be anticipated and with correct
chlorine residuals gives a clean surface with little or no change in
carrying capacity of the main. It is a brittle material which can fail
catastrophically. Maximum pipe deflection (change in diameter/original
diameter) should be 2% to avoid risk of cracking. Failure of lining has
also been attributed to pressure transients.
1.9.5 Paint systems

Often found on raw water lines, for example wellfield risers, paints are
applied either directly onto steel or onto an existing cement mortar
lining. The paint system may be used to inhibit corrosion of a cement
mortar lining in a low pH water environment.
1.9.6 Polyethylene lining

Heated steel pipes are dipped into a bath of powdered polyethylene thus
providing an external coating and a lining in one operation. May be
used on sewers carrying aggressive contents and also on risers of well-
fields.
1.10 Plastic pipes

Plastic pipes represent an area of growth. All plastics are corrosion
resistant, ductile and relatively impervious. Generally no protective
lining is applied either internally or externally. These materials can be
subdivided into thermoplastics and glass-fibre-reinforced thermosetting
resins. Both categories are available in a range of synthetic materials,
giving a wide-range spectrum of chemical resistance.
1.10.1 Thermosetting plastics

Thermosets are hardened by heating usually in the presence of a
catalyst. The hardened state is retained under reheating.
In general, thermoset pipes have greater stiffness, lower thermal
expansion coefficients and can be used at higher temperatures than
thermoplastics. Some thermoplastics may be tougher but not stronger
than thermosets which include glass-reinforced plastic (GRP), fibre-
reinforced plastic (FRP) and reinforced plastic matrix (RPM).
Thermoset pipes are laminates in which a relatively brittle but
chemically resistant resin is reinforced using strong and stiff glass
14
fibres which themselves have very poor chemical resistance in an

aqueous environment. Heterogeneous materials used to form these
fibre-reinforced pipes tend to be tougher than thermoplastics and the
energy required to produce a crack is greater. While fast cracks are
possible where only short discontinuous fibres are used as reinforce-
ment, failure in most cases is preceded by leakage due to cracking of
the resin between fibres, producing a flow path for pipe contents to
leak. To guard against this happening a 0.2% strain limit is recom-
mended.
Pipes made from thermosetting materials are usually classified in
terms of pressure rating up to around 64 bar g and ring stiffness EI=D3
from 250 N/m2 to 8000 N/m2 .
1.10.2 Thermoplastics
Thermoplastics are synthetic materials softened by application of heat
and are capable of repeated softening by subsequent heating. The
principal types of thermoplastics used in the water industry are polyvinyl
chloride (PVC) and polyethylene (PE). Other materials such as acrylo-
nitride butadience styrene (ABS) and polypropylene (PP) may be used
in special circumstances.
PVC was developed by German scientists shortly before World War
II. Meanwhile in the UK, ICI scientists had discovered a means of
producing polyethylene. Being lightweight, flexible and virtually free
from chemical attack made these materials contenders for pipeline
materials.
Initial poor performance of uPVC was attributed to a combination of
poor installation, interference damage and unsuitable operating condi-
tions including high transient pressures. From about 1973, when the
industry fully appreciated the limitations of the material, the failure
rate of uPVC has been below that of spun grey iron (cast iron).
Pressure pipes for water services may be obtainable in four classes as
indicated in Table 1.6.
Table 1.6. Four classes of pressure pipes
Pressure class
B C D E
Pressure, bar g 6 9 12 15
15
Table 1.7. The five series for HDPE defined in DIN 8074
Series
1 2 3 4 5
Pressure, bar 2.5 3.2 4.0 6.0 10.0
Medium-density polyethylene (MDPE) has to some extent replaced

the earlier high-density polyethylene (HDPE) and low-density poly-
ethylene (LDPE). Materials are defined by density ‘’ as follows:
LDPE 0:910 0:925
MDPE 0:925 0:940
HDPE 0:940 0:965
There is also an ultra-high molecular weight, high-density polyethylene
(UHMW-HDPE). LDPE has found application in small-bore water
pipes while MDPE and HDPE may be used for water main and effluent
main duties including sea outfalls. UHMW-HDPE is used in the trans-
port of abrasive slurries. Normal operating temperature range is from
408C to þ608C. Jointing is usually by butt-fusion welding up to
D ¼ 1600 mm and extrusion welding for D > 1600 mm.
DIN standard 8074 defines 5 series for HDPE, as shown in Table 1.7.
Other standards for polyethylene have a classification system based
upon the diameter/wall thickness ratio D=s ¼ standard dimensional
ratio (SDR) and based on continuous operation for 50 years at 208C
(Table 1.8).
For liquids other than water these pressures may be reduced by a
chemical resistance factor. At temperatures greater than 208C the
maximum continuous working pressure reduces so that by 608C,
maximum pressure is only 40% of the value at 208C.
Table 1.8. Classification system for PE
Type of Maximum operating pressure, bar g

material
Water Industrial
SDR11 SDR17 SDR11 SDR17
PE 80 12.5 8.0 10.0 6.0

PE 100 16.0 10.0 16.0 10.0
16
Table 1.9. Wall thickness and pressure ratings for PP
Class
A B C D E
Pressure, bar 3 6 9 12 15
Polypropylene (PP) may be used for higher-temperature applications,

e.g. normal service temperatures up to 808C and even 1008C with a
reduced life. The same jointing techniques apply as for a polyethylene
pipe.
Specifications usually define wall thickness for a standard series of
pressure ratings, e.g. for PP (Table 1.9). ‘Alloys’ of thermoplastics
have also been used in pipe production. An example of this is the
Hepworth Industrial Plastics range of Hep3O mains which is
produced from three materials: chlorinated polyethylene (PE), poly-
vinyl chloride (uPVC) and acrylic derivatives. The objective is to
combine the tough ductile character of PE with the higher strength
of uPVC.
Composites of both thermoplastics and thermosets have been used,
such as Permastran which is made up of a PVC core overlaid with
continuous rovings of fibreglass bonded with epoxy resin. The fibreglass
fibre provides Permastran with its hydrostatic strength while the PVC
provides a corrosion-resistant core. For a pipeline designed for
maximum working pressure of 350 psi a safety factor of 2 would be
provided and a maximum of 350 þ 42 psi would be allowed for surge
pressures.
1.11 Failure modes of pipes

An understanding of pipe strength and deformation limits is necessary
to prevent premature failure. Two main forms of failure are collapse and
brittle failure. Brittle fracture is a phenomenon in which a long, fast
brittle crack runs along the pipeline in a characteristically wavy
manner. Susceptibility of thermoplastics to this is dependent upon
material and principally hoop stress. Brittle fracture in uPVC follows
the slow growth of a crack from a defect or region of different mechan-
ical properties. A conservative 0.5% hoop strain limit will avoid brittle
fracture over a 50-year life.
17
1.12 Maximum pressure and allowable amplitude of surge in

plastic pipes
Unlike other materials, plastic pipes generally have no overpressure
allowance and the maximum sustained allowable pressure is also the
peak transient pressure. In addition some thermoplastic pipes have a
restriction placed upon the amplitude of pressure change experienced
during a transient event where this can be considered detrimental to
the life expectancy of the pipe. Typically for PVC and PE this amplitude
restriction has been set at half maximum sustained working pressure.
Since thermoplastic pipe materials have a lower strength than other
materials such as DI and steel, thicker walls are used to compensate.
The diameter of pipe available at higher pressure will also be more
limited. The typical life of these pipes has been set at 50 years by manu-
facturers although this is conservative.
1.13 Minimum pressures

As far as low or underpressures are concerned matters can be more
complicated. First there is the matter of possible damage to the pipe
to be considered. Thin-walled pipes, especially those having a relatively
small deformation modulus, may be at risk of collapse due to buckling as
a consequence of a lower pressure within the conduit than that external
to the pipe.
Fatigue damage may also occur and ovalization of a flexible pipe can
also cause high stresses in the pipe wall. For flexible conduits the shape
of cross-section may deviate from an original circular form. This process
of ovalization can increase over time and with falling pressure resulting
in reduced pipe stiffness. This aspect is considered in more detail in
Chapter 22.
Under sub-atmospheric conditions gas can evolve from solution with
potentially adverse consequences. Collapse of gas- or vapour-filled
cavities may possibly result in overpressures. It has also been reported
by Glass (1980) that evolution of certain gases from solution in
contaminated water could be responsible for promoting attack by
sulphate-reducing bacteria on the interior of a bitumen-lined steel
pipe in the same way as anaerobic corrosion is known to affect the
exterior of pipes.
A second consideration relates to the nature of the liquid being
conveyed. Codes of practice recommend that where treated water is
being carried then positive pressures should be maintained within the
pipeline system to avoid risk of contamination. Piezometric level
18
within the pipeline should be above possible groundwater levels and

also above the operating levels of any air valves which may be
present. By maintaining internal pipeline pressure at or above these
levels any leakage will be from inside the system to outside, thus
avoiding the possibility of contamination by groundwater. By keeping
minimum piezometric height above the operating level of any air
valve, the possibility of ingress of contaminants via the valve is also
avoided. A minimum pressure head of around þ2 m water gauge
(mWG) measured from the crown of the pipe may be an appropriate
minimum to adopt in many circumstances. Where the pipe is deeply
buried, this minimum may be raised, again to ensure that internal
pressure always exceeds external pressure from groundwater.
1.14 When is analysis necessary?

The questions are sometimes asked, ‘under what circumstances is an
analysis required and is there a convenient checklist which can be
applied?’ Each pipeline system is typically unique, having its own
characteristics defined by topography, flows and other factors. It is not
possible to be definite in saying that a specific scheme does not require
any analysis. Caution would suggest that some form of study be carried
out, even if not a detailed computer analysis. In the past, guidelines
have been laid down to identify these pipeline systems at risk and in
need of detailed examination. One such is the ASCE (1975) two-stage
checklist. Looking through the categories of pipeline included one might
get the impression that the large majority of systems require investigation.
Rather than essentially duplicating the list provided by the American
Society of Civil Engineers, some comments will be made. The increased
availability of programs and their ease of use, allow a pipeline to be
studied quite quickly.
Only for very straightforward systems can an analysis be avoided. For
instance, a short, simple line operating under low head and with rela-
tively low velocities might be exempt from an initial study, except
where the issue of sub-atmospheric pressures could be problematic,
i.e. risk of pipe collapse. During commissioning there is an opportunity
to measure transient pressures and introduce any protection which may
be required. Increasing concern over water quality and the avoidance of
vacuum pressures has made it necessary to study pipeline systems where
simply pipe strength considerations might have made the need for
analysis less important. The earlier checklist did not include liquid
quality considerations.
19
Longer pipelines, parallel lines of differing characteristics, branching

and looped systems should all be studied. Pipelines containing inter-
mediate features such as valves, booster pumps and tanks require
investigation. Particular attention should be paid to higher head
systems or schemes where velocities are large. Where air valve opera-
tion is possible or fast-acting valves are present, study of behaviour is
essential. Where water quality is a factor, analysis is also considered
necessary.
20
2
Derivation of basic equations
Two basic methods of analysis have been used extensively in the inves-
tigation of unsteady flows in pressure conduit networks. The simplest of
these has the assumption that the flowing liquid is incompressible and
that the conduit walls are rigid or inelastic. This method is sometimes
known as the rigid-column analogy and has been widely employed in
the analysis of longer-period oscillations such as occur in low-pressure
tunnels of hydropower networks.
The second approach considers the liquid to be compressible and the
conduit walls to be deformable resulting in a variation of flow along the
pipeline or tunnel at any instant. This ‘elastic’ approach coupled with
the large computing power and memory of modern desktop computers
has the potential to reproduce both high-frequency and longer-period
aspects of hydraulic transients in a way that was hitherto not possible.
2.1 The rigid-column approach

In this approach, as generally with the elastic analysis, it is assumed that
the head loss relationships that apply under steady flow conditions are
also valid during a transient event. Resistance laws for pipe flow can
thus be used to represent head loss along a main. Representing a
liquid-filled pipeline as a rigid body means that any change in velocity
imposed at one end of the line is instantaneously experienced
throughout the main.
An equation describing the relationship between flow and head can
be obtained by considering the elementary system of Fig. 2.1. A simple
pipeline is shown of diameter D, cross-sectional area A and length L.
Liquid of density flows into the pipeline from an upstream reservoir
passing through a valve at the downstream end of the main. Consider
21
Inertial head gradient
Hi
Static head line

M
hf
Resistance gradient
Valve closing
Diameter = D M
Velocity = V
Cross-section A
Fig. 2.1. Definition sketch for rigid column analysis
the valve to be closed at some rate that caused flow to decelerate at

a rate dV=dt and producing an inertial head rise Hi at the upstream
valve face. Forces acting upon the body of liquid contained in the
pipeline are:
Pipeline resistance ¼ gAhf ¼ gAfLV 2 =ð2gDÞ
Force imposed at the valve face ¼ gAHi
Inertia force ¼ AL dV=dt
where g is acceleration due to gravity and f is a friction factor.
Summing these forces and dividing throughout by gA gives:
fLV 2 =ð2gDÞ þ L=g dV=dt þ Hi ¼ 0 ð2:1Þ
Neglecting pipeline resistance produces the simple relationship:
Hi ¼ L=g dV=dt ð2:2Þ
This simple expression can be used to obtain a very rapid assessment of
transient pressure rise/fall for a given rate of flow change. Supposing a
valve closure at the downstream end of the pipeline occurs. As soon
as the valve has become completely closed then V ¼ 0 and
dV=dt ¼ 0; that is, no further change in velocity occurs. The inertial
head Hi becomes zero and the transient event is over. The limitation
of this approach arises when the magnitude of dV=dt is large and/or
the pipeline length L is considerable. Then values of Hi can become
unrealistically large and it is at this stage that it is necessary to
employ the elastic approach in order to obtain more realistic answers.
The incompressible flow equation (2.1) has another important function
22
in that it can be used to provide preliminary estimates for capacity of

some types of surge protection equipment and some examples of its
use in this regard will be considered later in this book.
2.2 Compressible flow theory

A straightforward derivation of the equations of motion in compressible
flow is presented in this chapter. Basically there are two independant
variables: x, the distance along the axis of a pipeline; and t, the time
which has elapsed from some starting-point. The distance along a pipe-
line may be þve or ve depending upon how the user has decided to
schematise a network but the normal flow direction in a line would
be taken as the þve direction. The time elapsed may also be fairly
arbitrarily assigned. For instance time t ¼ 0 might be taken at or
shortly before a transient event commences.
Dependent variables are fundamentally those of flow rate Q and
pressure p. These vary with both x and t. Assuming a uniform velocity
V over any pipe cross-section A then velocity can replace flow rate
through V ¼ Q=A. The relationships governing these variables are
those of conservation of mass and conservation of force and
momentum. Other variables such as air volume, water level or possibly
pump speed are related to the primary variables V and p through
additional equations.
2.2.1 Conservation of force

Consider a section of pipeline having an arbitrary shape of cross-section
A and of length dx as illustrated in Fig. 2.2. At any instant the forces
acting upon the body of liquid within this element are:
the net pressure force ¼ @p=@x dx A
the pipeline resistance force ¼ S dx gA
the axial component of liquid weight ¼ @z=@x dx gA
and the inertia force ¼ A dxð@V=@t þ V @V=@xÞ
where S is the rate of head loss due to pipeline resistance, is liquid
density, g is acceleration due to gravity and z is elevation of the pipe
centreline.
Summating these forces:
@p=@x dx A S dx gA @z=@xgA
A dxð@V=@t þ V @V=@xÞ ¼ 0
23
Piezometric line
p/(rg) + ∂[p/(rg)]/∂x dx
h = p/(rg)
rAV
H dx
rAV + ∂(rAV )/∂x dx

S
z z + ∂z/∂x dx
Horizontal datum
Fig. 2.2. Definition sketch for compressible flow analysis
and dividing throughout by A dx gives:

1= @p=@x þ gS þ g @z=@x þ @V=@t þ V @V=@x ¼ 0 ð2:3Þ
2.2.2 Conservation of mass

Considering the mass balance within the element over an increment of
time, dt:
the net mass influx to the element (neglecting second-order effects)
¼ A @V=@x dx V @A=@x dx VA @=@x dx
the rate of change of mass within the element
¼ @=@tðA dxÞ ¼ @A=@t dx þ A @=@t dx
24
Equating net mass inflow to rate of change of mass within the

element yields:
A @V=@x dx þ V @A=@x dx þ VA @=@x dx þ @A=@t dx

þ A @=@t dx ¼ 0
and dividing throughout by the mass of fluid within the element A dx

gives:
@V=@x þ 1=ð@=@t þ V @=@xÞ

þ 1=Að@A=@t þ V @A=@xÞ ¼ 0 ð2:4Þ
2.2.3 Compressible flow equations in terms of total head H

Introducing pressure head h ¼ p=ðgÞ and total head above datum
H ¼ h þ z then the force balance equation:
@V=@t þ V @V=@x þ 1= @p=@x þ gS þ g @z=@x ¼ 0 ð2:3Þ
becomes:
@V=@t þ V @V=@x þ 1=ðgÞð@H=@x @z=@xÞ þ gS þ g @z=@x ¼ 0
or finally:
@V=@t þ V @V=@x þ g @H=@x þ gS ¼ 0 ð2:5Þ
In the mass balance equation:
@V=@x þ 1=ð@=@p @p=@t þ V @=@p @p=@xÞ
þ 1=Að@A=@p @p=@t þ V @A=@p @p=@xÞ ¼ 0 ð2:2Þ
Since and A are taken to be solely dependant on pressure p, therefore,

d=dp ¼ @=@p and dA=dp ¼ @A=@p. Also dz=dx ¼ @z=@x since @z=@t
is assumed ¼ 0.
Substituting gðH zÞ for pressure p then:
@V=@x þ f1= d=dp þ 1=A dA=dpggf@H=@t þ V @H=@xg
f1= d=dp þ 1=A dA=dpgg dz=dx ¼ 0
Writing,
1=a2 ¼ ð1= d=dp þ 1=A dA=dpÞ
25
or,
p
a ¼ 1= fð1= d=dp þ 1=A dA=dpÞg ð2:6Þ
then the mass balance equation finally becomes:
a @V=@x þ g=af@H=@t þ V @H=@xg g=aV dz=dx ¼ 0 ð2:7Þ
26
3
Interpretation of a
A critical element of the equation of conservation of mass (equation

(2.7)) derived in Chapter 2 is the quantity a which is a complicated
function involving both the properties of the fluid and of the conduit.
The value assigned to this parameter will have a crucial role to play
in the prediction of pressure extremes. This chapter attempts to
describe some of the factors which should be considered when choosing
a value and also some of the uncertainties present in estimation.
In Chapter 2 the quantity a was established as:
p
a ¼ 1= fð1= d=dp þ 1=A dA=dpÞg ð2:6Þ
The value of a is dependent upon the properties of the fluid and d=dp
and upon the cross-sectional area of the conduit and its deformation
characteristics A and dA=dp.
3.1 Fluid properties

Considering fluid characteristics. For water in the liquid state, the bulk
modulus K is defined by:
K ¼ dp=ðd=Þ or 1=K ¼ 1= d=dp ð3:1Þ
For water, the value of K varies only slowly with temperature with a
decrease of the order of 2% between 08C and 508C. A value of
2.137 GN/m2 at 218C is given by Knapp et al. (1970) with Thorley
and Enever (1979) providing a slightly higher value of 2.193 GN/m2
at 208C. Seawater has a value of 2.4 GN/m2 at 258C. For a computer
simulation conducted assuming an almost uniform temperature, K
can be considered constant and independent of flow conditions. If it
27
is assumed that the body of water is infinitely large then:

p p
a ¼ 1= fð1= d=dp þ 1=1 dA=dpÞg or a ¼ fK=g
ð3:2Þ
Equation (3.2) is the speed of propagation of sound or of a pressure

wave through an unconstrained body of water. Equation (2.6) repre-
sents the speed of the pressure wave or sound wave through a fluid
body which is contained within a conduit.
3.2 Influence of the conduit wall

Introduction of the term 1=A @A=@p represents the influence of the
conduit containing the water. The simplest representation of the
pipeline’s effect on a is that for a relatively thin-walled circular elastic
pipe which is unconstrained in the axial direction. With reference to
Fig. 3.1, if a change of internal pressure dp takes place, then for
equilibrium:
dpD ¼ d2s or d ¼ D dp=ð2sÞ
Wall thickness = s Wall thickness = s
Diameter = D
ds
Change of internal
pressure = dp
Change of hoop stress = ds
Fig. 3.1. Forces at a pipe cross-section
28
Interpretation of a
also
Youngs modulus E ¼ d=ðdc=cÞ ¼ dpD=ð2s dD=DÞ
¼ D2 =ð2sÞ dp=dD
assuming that the pipe is entirely free to contract in length. Re-
arranging,
dD=dp ¼ D2 =ð2sEÞ
Cross-sectional area A ¼ =4D2 and A þ dA ¼ =4ðD þ dDÞ2 , or
neglecting second-order effects:
dA ¼ =2D dD or dA=dD ¼ =2D
and
1=A dA=dD ¼ 2=D
then,
1=A dA=dp ¼ 1=A dA=dD dD=dp ¼ 2=D D2 =ð2sEÞ
¼ D=ðsEÞ ð3:3Þ
3.3 Simple expression for a

Substitution of the relationships of equations (3.1) and (3.3) in the
expression for a then:
p
a ¼ 1= f½1=K þ D=ðsEÞg
To accommodate these conditions in which the pipe is not entirely
free to contract longitudinally due to restraint, a factor c1 is introduced
so that finally:
p
a ¼ 1= fð1=K þ Dc1 =ðsEÞÞg or
p p
a ¼ ðK=Þ= ð1 þ DK=ðsEÞc1 Þ ð3:4Þ
The speed of propagation of the pressure wave through an uncon-
fined body of liquid is modified by the effect of the containing pipe
walls.p In all cases a reduction in the unconfined value for
a ¼ ðK=Þ is experienced. Liquid and pipe are considered to act as
a composite medium giving a single value of a for this two-component
arrangement. This is justified by the continuous coupling action
between liquid and conduit throughout the system. Calculations can
be carried out for a pipeline system modelled with the liquid and
29
pipewall represented as separate components using individual propaga-

tion rates for liquid and conduit but these are usually confined to the
analysis of vibrations. Some examples of these types of analyses are
described in papers by Ellis (1980), Kot et al. (1980), Wilkinson
(1980) and Heinsbroek and Tijsseling (1994). These methods are not
used in day-to-day pressure transient investigations.
For the simple case considered, wave speed a is seen from equation
(3.4) to be a constant for each pipeline of a system, independent of
flow conditions. Individual values of a used in analyses can vary consid-
erably from case to case depending upon the geometrical characteristics
of the pipe and its material. Figure 3.2 displays the possible range in
wave speed as a function of elastic modulus E and the ratio Dc1 =s for
Steel
a = 1350 1300 1200 1100 1000 900 800 700 600
Hard cast iron
500
Soft cast iron 400

11
300
Concrete
Asbestos cement
200
Polyester
(glass fabric)
Polyester 10
(glass fibre)
100
Perspex
PVC (hard)
PVC 50
(blow resisting)
PE (hard) 9
PE (soft)
log10 (E )
8 1 2 3
log10 (Dc 1/s)
a = ÷(K/r)/÷[1 + DKc 1/(sE)]
Fig. 3.2. Acoustic wave speed a as a function of E and Dc1 /s
30
Interpretation of a
a circular pipe section. A wave speed value for wood stave pipe was
given by Fedosoff and Szpak (1979) as 762.0 m/s. The influence of
constraints has been considered by various authors such as Streeter
and Wylie (1967), Thorley and Enever (1979) and Jaeger (1977).
Expressions for the constraint factor c1 for different circumstances,
including thick-walled conduits defined by D=s < 25, and tunnels
have been included in Table 3.1.
Table 3.1. Constraint factor c1 for various conditions

p p
Pipelines: a ¼ ðK=Þ= ð1 þ D=s K=E c1 Þ
Thin-walled: D=s 25 Thick-walled: D=s < 25
Anchored at one end and
other end free
c1 ¼ 5=4 c1 ¼ D=ðD þ sÞð5=4 Þ þ 2s=Dð1 þ Þ
Anchored at both ends
c1 ¼ 1 2 c1 ¼ D=ðD þ sÞð1 2 Þ þ 2s=Dð1 þ Þ
Expansion joints
c1 ¼ 1 =2
Expansion joints
throughout
c1 ¼ 1 c1 ¼ D=ðD þ sÞ þ 2s=Dð1 þ Þ
c1 is usually in the range 0:85 c1 1:25

p p
Tunnels: a ¼ ðK=Þ= ð1 þ K=Ec1 Þ
Unlined: E ¼ Er
c1 ¼ 2ð1 þ Þ
Lined: E ¼ lining modulus, Er ¼ rock modulus
c1 ¼ 1=ð1 þ 12 D=s Er =EÞ
31
3.4 Variation of a with conduit shape

In the case of other cross-sectional shapes such as square ducts for
instance, matters are complicated by deformation from the original
shape under atmospheric conditions, as pressure is increased or
decreased. Thorley and Enever (1979) provide extensive information
on propagation speeds for a range of conduit shapes. The wave speed
remains constant for a given shape of liquid-filled pipe and a specific
value of D=s. They show a general equation for polygonal ducts:
1=A dA=dp ¼ 1=EfD=s þ tan4 ðD=sÞ3 =15g
þ tan2 ðD=sÞ=ð2GÞ ð3:5Þ
where is the angle between flats, D is the distance between the flats, s
is wall thickness and G is the shear modulus of pipe wall material. It will
be noted that as the number of flats approaches 1 the shape becomes
increasingly circular, tends to zero and the expression 1=A dA=dp
approaches D=ðsEÞ.
3.5 Influence of gas on a

Thus far the composite wave speed a has remained independent of flow
conditions. Gases are present in the majority of liquids, both as free gas
in the form of bubbles and also as dissolved gas. As pressure is increased,
the amount of gas which can be dissolved also increases, thus reducing
the initial amount of free gas. According to Knapp et al. (1970),
dissolved gas has a negligible effect upon the tensile strength of a
liquid and thus will not significantly influence the acoustic wave
speed. However, the presence of free or undissolved air or gas within
the liquid body introduces a further component to the composite
medium which does have an important influence upon speed of pressure
waves. Deformation of the liquid—gas mixture with changing pressure is
greatly influenced by the much greater compressibility dg =dp of the gas
with g representing the gas density.
If a mixture of gases is in contact with the liquid surface, Henry’s law
states that the amount of each gas that will dissolve at equilibrium is
governed by its partial pressure. This can be written as: mg ¼ ps ,
where mg is the mass of gas dissolved/unit volume of liquid, is a
temperature-dependent solubility constant and ps is equilibrium
partial gas pressure (bar). Under conditions just described, maximum
or saturation concentration of air in water at normal temperature and
atmospheric pressure is just under 2%, of which one-third is oxygen
32
Interpretation of a
and two-thirds is nitrogen. Although its solubility is high, the amount of

CO2 dissolved is very low since air contains only 0.03% of this gas. The
maximum air content < saturated concentration unless special treat-
ment has been carried out.
If pressure should fall, the liquid becomes supersaturated with gas for
that pressure and so some dissolved gas is released from solution
although it usually takes several seconds for a new equilibrium to be
reached. During a short-lived transient event equilibrium may not be
achieved. Gas release rate is greater than the corresponding gas absorp-
tion rate and so there is a net amount of free gas in the form of bubbles.
This release of gas is termed ‘gaseous cavitation’. If pressure drops below
the saturation pressure of liquid vapour, a very fast release occurs
termed vaprous cavitation within a few micro-seconds.
Expressing dissolved gas mass as an equivalent volume of gas Volg
under atmospheric pressure and at 08C, the following relationship is
obtained:
Volg =VolL ¼ ps =1:013
where VolL is liquid volume, is a solubility constant in units of volu-
metric concentration ¼ 0.0187 for air at 208C. Void fraction 2 can be
written:
2 ¼ Volg =ðVolg þ VolL Þ
provided no slippage is assumed between bubbles and flowing liquid.
The notation 2 is proportional to absolute pressure. If saturated void
fraction 2 is about 2% at atmospheric pressure patm , then at
1
2 patm 2 ¼ 1% with the other 1% having come out of solution and
become free gas. As pressure drops, free bubbles of gas expand following
the gas law, aided in their growth by gas evolving from the now super-
saturated liquid and diffusing into the bubbles. Eventually a bubble
reaches a critical diameter when the differential pressure across the
bubble wall exceeds the surface tension force. Explosive growth in
bubble size then occurs with bubbles starting to rise towards the
crown of the pipe. Since the void fraction was related to a specific pres-
sure, usually atmospheric pressure, any gas which comes out of solution
at a lower pressure will have a volume given by the gas law:
pabs Volf ¼ constant
where is the ratio of specific heats for the gas and Volf is the amount of
free or undissolved gas. Then, if the amount of gas coming out of
solution is f and the initial amount of free gas is 1 , both being
33
measured at atmospheric pressure patm , then at some pressure pabs the

volume fraction will be:
¼ ð1 þ f Þðpatm =pabs Þð1=Þ
It takes a finite time for a new equilibrium condition to become estab-
lished and so the amount of gas evolving from solution f will not
immediately be the total amount for the pressure pabs . The rate at
which gas is released from solution is dependent upon a number of
factors. Dissolved gas can only appear from solution if a ‘free phase’
boundary exists. This requires the presence of interstices in conduit
walls where free gas can lodge and/or particles within the liquid
which can act as ‘hosts’ for bubbles of free gas. Crevices and small
cracks in pipe walls are found, also micro-nuclei within the flow.
Keller and Zielke (1976) carried out filtering and rinsing of test
apparatus with filter size 0.5 mm. Despite this they found an average
of 2.8 nuclei/cm2 with particles as large as 120 mm present. It is a
practical impossibility to completely clean pipes and tunnels of dust
particles. Turbulence can also influence the rates of gas release from
solution and some observations have shown that as Reynold’s
number (Rn) increases, the rate of release also rises and the time for
bubble content to stabilise falls. For an Rn of 370 000, bubble content
stabilises after 9 s whereas if Rn ¼ 590 000, bubble content is stabilised
after 5—6 s. Since the gas release rates depend on a range of factors, it is
not surprising to find alternative approaches being used by different
modellers. For example, Fox (1977) assumed that all dissolved gas
evolved from solution with no delay and that the liquid was initially
saturated with 2% gas at atmospheric pressure. Other investigators
have adopted an exponential gas release rate. What has been noted
is that including some air produces essentially the same result and
predictions were not very sensitive to the initial free gas content,
release rate or amount of gas released.
To illustrate the influence of free gas upon behaviour of the pressure
wave consider a liquid having total void fraction at the prevailing
absolute pressure pabs . Let the gas density be g with liquid density .
If the volume of free gas is Volf and volume of liquid is VolL , total
volume Vol ¼ Volf þ VolL . Mean density of the mixture m is given by:
m ¼ ðg Volf þ VolL Þ=Vol
or
m ¼ g þ ð1 Þ
34
Interpretation of a
Compressibility of liquid is represented by:

1=K ¼ 1= d=dp
and compressibility of gas by:
1=Kg ¼ 1=g dg =dp
Compressibility of the mixture can then be represented by:
=Kg þ ð1 Þ=K
Substituting in the expression for wave speed a then for a circular pipe
section:
p
a ¼ 1= fm ½Dc1 =ðsEÞ þ =Kg þ ð1 Þ=Kg ð3:6aÞ
Essentially the same equation has appeared in a number of guises.
Martin et al. (1976) assumed the contribution of the gas fraction to
mixture density was small and so:
m ð1 Þ
giving
p
a ¼ 1= fð1 Þ½Dc1 =ðsEÞ þ =Kg þ ð1 Þ=Kg ð3:6bÞ
Tullis et al. (1976) assumed void fraction to be small giving
1 1 so that for a thin-walled pipe free to contract in length:
p
a ¼ 1= fðD=ðsEÞ þ =Kg þ 1=KÞg ð3:6cÞ
The compressibility of gas represented by Kg can be replaced as follows.
Applying the gas law to a fixed mass of free gas then:
pabs Volf ¼ constant
where ¼ Cp =Cv is the ratio of specific heats for the gas, Cp is specific
heat at constant pressure, and Cv is specific heat at constant volume.
All diatomic molecules of gas display the same value of ¼ 1:4 at
moderate temperatures applicable for pipeline transients. Texts on
thermodynamics will yield additional details, for example Spalding
and Cole (1963). Differentiating with respect to Volf then,
ð1 þ Þ
dpabs =dVolf ¼ constant=Volf
or
dpabs =pabs ¼ dVolf =Volf
Also the mass of gas is given by:
g Volf ¼ constant
35
Again, differentiating with respect to Volf ,

dg =dVolf ¼ constant=Vol2f ¼ g =Volf
and rearranging,
dg =g ¼ dVolf =Volf
therefore,
dpabs =pabs ¼ dg =g
Gas compressibility is thus represented by:
1=Kg ¼ 1=g dg =dpabs ¼ 1=ðpabs Þ ðdpabs =dpabs Þ ¼ 1=ðpabs Þ
Making these substitutions the equation for wave speed is then:
p
a ¼ 1= f½ð1 Þ þ g ½Dc1 =ðsEÞ þ =ðpabs Þ
þ ð1 Þ=Kg ð3:7Þ
Fox (1977) neglected the gas mass contribution to mean density,
giving, m ð1 Þ and assumed 1 1 in the liquid compressi-
bility term, also ¼ 1:0, thus yielding the simpler expression:
p
a ¼ 1= fð1 Þ½D=ðsEÞ þ =pabs þ 1=Kg ð3:8Þ
This form of the equation for a perhaps highlights the fact that wave
speed is dependent upon absolute pressure pabs or has become a
function of the prevailing transient flow condition where free gas is
considered as a constituent of the fluid in the pipeline. Without the
influence of gas, the wave speed remains constant and independent
of flow conditions for a specified pipeline.
Acoustic velocity a is strongly dependent on void fraction . Even a
small amount of free gas produces a dramatic fall in wave speed. At the
opposite end of the scale, as void fraction tends to 100%, the wave
speed rises towards the speed of sound in air given by:
p
a ¼ fpabs =g g
Adopting standard values at atmospheric pressure and ambient
temperature, g ¼ 1:22506 kg=m3 , patm ¼ 101:3 kN=m2 and ¼ 1:4,
then the speed of sound in air is calculated as 340 m/s.
The gas release head is considered to be around 2.4 m absolute (Fox,
1977). Once the gas has evolved from solution it is more difficult to
force it back into solution by a simple rise to atmospheric pressure
and requires a more substantial pressure increase before much of the
gas will be forced to redissolve. Fox proposed that all gas came out of
36
Interpretation of a
1200
1000
800
Wave speed (m/s)
600
Vf = 1%
400 Vf = 0.1%
Vf = 0.01%
Vf = 0.001%
Vf = 0.0001%
200
0
0.1
0.5
0.9
1.3
1.7
2.1
2.5
2.9
3.3
3.7
4.1
4.5
4.9
5.3
5.7
6.1
6.5
6.9
7.3
7.7
8.1
8.5
8.9
9.3
9.7
10.1
Pressure (bar(abs))
Fig. 3.3. Acoustic wave speed plotted against pressure for various void fractions
solution at this gas release head and that equilibrium concentration was
achieved with no delay. Other modellers have adopted an exponential
decay of rate of gas release.
Figure 3.3 shows how wave speed changes with pressure for a range of
free gas concentrations. The gas fractions are those at atmospheric
pressure and range from 0.0001% to 1.0% by volume.
Adoption of a wave speed which varies with changing pressure
conditions within each pipeline of a network has important implications
with regard to the way in which computations are conducted. In the
presence of free gas the liquid in the pipeline system over most of the
range in free gas concentration has effectively become a fluid which
retains a density dominated by the weight of the liquid component
but with compressibility more like that of the gas. Only at the extremi-
ties of concentration range does this alter, with liquid properties
dominant when only very small amounts of gas are present, and gas
properties dominant if only quite small amounts of liquid are available.
Evolution of gas from solution and subsequent expansion of the free
gas has the effect of introducing an additional source of fluid to the
system. Suppose that upstream of a region in which free gas is present
and expanding, the flow is decelerating at some rate dV=dt. Down-
stream of the zone of gas expansion the flow will be decelerating at a
lesser rate as the flow from upstream will be augmented by liquid
released from the length of pipe containing the expanding gas in
37
order to accommodate the increasing gas volume. The effect is to

reduce the downstream rate of deceleration and suppress to a significant
extent further pressure drop.
The equation for wave speed a in the presence of a free gas consti-
tuent has been developed on the basis of a homogeneous dispersal of
gas. Observations have revealed a large number of gas/liquid flow
regimes with no standard definition for each type of flow. Escarameia
(2005) mentions that up to 84 different terms have been in use and
illustrates some of the flow patterns for both horizontal and vertical
movement.
The present analysis has neglected ‘slip’ velocity between gas and
liquid and this can have significant implications in some circumstances.
Consider a descending section of pipeline carrying a range of bubble
sizes. The larger bubbles will tend to move more slowly — that is,
have a larger slip velocity than smaller bubbles. The smaller bubbles
will catch up and be absorbed by larger bubbles, resulting in still
larger and slower-moving bubbles. The large bubble may eventually
come to rest and start to move back upstream, collecting yet more
small bubbles and causing variations of void fraction not accommodated
in the simple theory presented. This type of behaviour can also result in
a pulsating type of flow behaviour.
A further departure arises when a gas/liquid regime such as the ‘slug’
flow approaches a local source of head loss such as a valve. The loss in
pressure, p, across the valve is given by:
p=ðm gÞ ¼ KL V 2 =ð2gÞ
When the mixture is predominantly liquid, m will be close to the

liquid density and p will be relatively large. If a ‘slug’ of gas should
arrive at the valve, the mixture density will fall and pressure drop p
will also drop proportionately, causing a pressure wave to propagate
from the valve into the pipeline. This process will occur each time a
‘slug’ of gas passes through the valve. This phenomenon is not accom-
modated by the straightforward theory.
3.6 The effect of sewage

Sewage is a generic term including the following:
. soil or sanitary sewage from households
. industrial wastewaters
. cooling water or ‘thermally’ contaminated water
38
Interpretation of a
. surface water from rainfall and storm water

. drainage water from building sites, fields and leakage from broken
pipes.
The temperature of sanitary sewage tends to be relatively uniform all
year round, usually between 108C and 218C. From 300 mg/l to 500 mg/l
of dissolved solids and 200 mg/l to 400 mg/l of suspended solids
including grease will have been added. Septic sanitary sewage may
under some circumstances produce hydrogen sulphide, which can be
corrosive to cement products such as concrete and asbestos cement
pipes. Industrial wastewater can vary considerably in temperature and
other physical characteristics and ranges from strongly acid to strongly
alkaline, requiring special materials for pipelines, pumps and fittings.
Dissolved and suspended solids may conceivably affect the density
and viscosity of the fluid. Where solid particles or larger aggregations
of matter are present in sewage this adds a further component to the
medium. Depending upon the density and compressibility of this
material carried along in the liquid there may be an effect upon the
acoustic wave speed. Generally, with both a free gas component and
solid matter present, then the average density of the medium will be:
m ¼ ðVolL þ g Volg þ s Vols Þ=ðVolL þ Volg þ Vols Þ
if VolL þ Volf þ Vols ¼ Volt , Volf =Volt ¼ and Vols =Volt ¼ s then:
m ¼ ð1 s Þ þ g þ s s
Average compressibility becomes:
ð1 s Þ= d=dp þ g =g dg =dp þ s =s ds =dp
¼ ð1 s Þ=K þ =Kg þ s =Ks
¼ ð1 s Þ=K þ =ðpabs Þ þ s =Ks
then
p
a ¼ 1= fm ½ð1 s Þ=K þ =ðpabs Þ
þ s =Ks þ Dc1 =ðsEÞg ð3:9Þ
The manner in which Ks changes with pressure will depend upon
constituents of the sewage. If compressibility is relatively low, like
water, then s =Ks constant.
Some effluent pipelines are designed to convey sewage of a particular
type, for example from a residential area, and the solids content may be
assessed on this basis. Other lines may carry mixtures having a wider
39
range of properties from domestic and industrial wastes to stormwater

flows. Properties of the mixture will vary significantly and it is recom-
mended that investigations should include properties of stormwater
when flows may well be at a maximum. Under dry weather conditions,
primarily domestic and/or industrial waste waters may be carried. While
mean daily flows may be relatively modest, these may be conveyed in
the form of intermittent ‘plug’ flows in which discharge rates are close
to the pumps’ duty point. It may well be necessary in such circum-
stances to investigate flows involving different mixture properties.
40
4
Characteristic equations
Chapter 3 discussed how the acoustic wave speed in a pressure pipeline

system can be influenced by the properties of both the pipeline and by
any inclusions in water being conveyed. As far as sewerage systems are
concerned, a large part of the function of the water is to carry contami-
nants and impurities to treatment or disposal and it is to be expected
that both gas and some solid constituents will be present. This
chapter uses the partial differential equations together with the acoustic
wave speed a, to develop what are known as characteristic equations.
4.1 Development of characteristic equations

The force balance equation was derived in Chapter 2 as:
@V=@t þ V @V=@x þ g @H=@x þ gS ¼ 0 ð2:5Þ
and the equation balancing mass flow and storage became:
a @V=@x þ g=af@H=@t þ V @H=@xg g=aV dz=dx ¼ 0 ð2:7Þ
It is possible to proceed directly to the solution of pressure transient
flow problems using the above equations. However, many existing
modelling tools employ what are known as the ‘characteristic’ equations
and a simple derivation of these equations from the basic force and
mass balance equations (2.5) and (2.7) follows. These characteristic
equations have been used throughout this book to illustrate various
aspects of pressure transient behaviour.
Adding equations (2.5) and (2.7) balancing force and mass flow then:
@V=@t þ ðV þ aÞ @V=@x þ g=að@H=@t þ ðV þ aÞ @H=@xÞ
þ gS g=aV dz=dx ¼ 0
41
and subtracting equation (2.7) from equation (2.5) then:

@V=@t þ ðV aÞ @V=@x g=að@H=@t þ ðV aÞ @H=@x þ gS
þ g=aV dz=dx ¼ 0
The derivatives @ð Þ=@t þ ðV aÞ @ð Þ=@x represent total rates of

change in the variables V and H along paths dx=dt ¼ V a in the
x—t plane and the equations can be written as ordinary differential
equations along these paths so that:
dV=dt þ g=a dH=dt þ gS g=aV dz=dx ¼ 0 ð4:1Þ
along dx=dt ¼ V þ a ð4:2Þ
and
dV=dt g=a dH=dt þ gS þ g=aV dz=dx ¼ 0 ð4:3Þ
along dx=dt ¼ V a ð4:4Þ
Multiplying throughout these equations by dt then:
dV þ g=a dH þ gS dt g=aV dz=dx dt ¼ 0 along dx=dt ¼ V þ a
and
dV g=a dH þ gS dt þ g=aV dz=dx dt ¼ 0 along dx=dt ¼ V a
Integrating the equations along the paths dx=dt ¼ V a then:
ð ð ð ð
dV þ g 1=a dH þ g S dt g V=½aðV þ aÞ dz ¼ constant
along dx=dt ¼ V þ a
and
ð ð ð ð
dV g 1=a dH þ g S dt þ g V=½aðV aÞ dz ¼ constant
along dx=dt ¼ V a
or if a is assumed constant:
ð ð
V þ g=aH ¼ constant g S dt þ g V=½aðV þ aÞ dz ð4:5aÞ
along dx=dt ¼ V þ a ð4:2Þ
42
and
ð ð
V g=aH ¼ constant g S dt g V=½aðV aÞ dz ð4:6bÞ
along dx=dt ¼ V a ð4:4Þ

Ð Ð
Neglecting the terms g S dt and g V=½aðV aÞ dz for the moment
gives the simple linear equations:
V þ g=aH ¼ constant ð4:5bÞ
along dx=dt ¼ V þ a
and
V g=aH ¼ constant ð4:6bÞ
along dx=dt ¼ V a
The paths dx=dt ¼ V a are ‘characteristics’ of the system. The
derivation of the ‘characteristic’ equations given above represents a
simple approach. There are more sophisticated and elegant derivations.
Characteristics are features of many engineering disciplines and for an
introduction to the application of the ‘Method of Characteristics’ the
book by Abbott (1966) is highly recommended.
The constant or ‘invariant’ quantity in each of the above equations
(4.5b) and (4.6b) is often named after Riemann, who was one of the
first to use what has become known as the ‘Method of Characteristics’.
He applied this method to the field of gas dynamics in the mid-19th
century. Abbott used the symbol J to denote this invariant quantity
so that, neglecting the integrals on the right-hand side of the equations:
V þ g=aH ¼ Jþ along dx=dt ¼ V þ a
and
V g=aH ¼ J along dx=dt ¼ V a
As written above, the invariant J has units of velocity. Alternatively,
the invariants could be in units of head by multiplying throughout by
a=g, giving:
H a=gV ¼ J
In this book J has been used in terms of velocity. There are no compel-
ling reasons for this choice other than that being in units of velocity the
quantities involved are typically small with possibly smaller risk of
misplacing the decimal point when carrying out manual calculations.
43
Also in other applications of characteristics such as to free surface flow

transients or to supercritical steady flow, equations are in terms of velo-
city without the option to convert to other units. It could also be
mentioned that in the presence of free air or gas a becomes a function
of H and so having the term g=aH seems appropriate.
4.2 Significance of the integrals

The pair of integrals on the right-hand
Ð side of the equations represent the
effects Ðof pipeline flow resistance g S dt and changes in pipeline eleva-
tion g V=½aðV aÞ dz. These terms still exist even under steady
flow. Including these effects renders the quantity J quasi-invariant with
the value changing gradually along the paths dx=dt ¼ V a.
The following sections consider each of these integrals in isolation.
4.3 Effect of changing pipe elevation

Neglecting pipeline resistance for the moment, along a path
dx=dt ¼ V a the equations for the quasi-invariants become:
ð
V g=aH ¼ J g V=½aðV aÞ dz ð4:5cÞ
Ignoring any small changes in kinetic energy head then in the absence
of pipeline resistance and losses generally H remains essentially
constant along a path dx=dt ¼ V a so that along some length of
conduit over which velocity changes by an amount dV and over
which pipeline elevation changes by an amount dz:
ð ð
dV ¼ g V=½aðV aÞ dz or since V a; dV g V=a2 dz
Considering steady motion, mass flow is constant, or: AV ¼ constant,

therefore,
ð þ dÞðA þ dAÞðV þ dVÞ ¼ AV
Neglecting second-order effects and dividing throughout by mass flow
rate:
d= þ dA=A þ dV=V ¼ 0
then,
1= d=dp þ 1=A dA=dp ¼ 1=V dV=dp
44
and substituting:
1=K þ Dc1 =ðsEÞ ¼ 1=V dV=dp
Since pressure change dp is due entirely to elevation change dz then
dp ¼ g dz, or,
½1=K þ Dc1 =ðsEÞ ¼ 1=V dV=ðg dzÞ
Substituting a2 and rearranging,
gV=a2 dz ¼ dV
Between limits 0 and 1 then:
ð
V1 V0 ¼ g V=a2 dz ð4:7Þ
The equation can also be integrated in steady flow to give,

g=a2 dz ¼ dV=V or; g=a2 ðz1 z0 Þ ¼ lnðV1 =V0 Þ
where symbol lnð Þ denotes the natural logarithm, so that finally,
V1 ¼ V0 exp½g=a2 ðz1 z0 Þ ð4:8Þ
The above demonstrates that this small integral represents the
velocity difference caused by compression or expansion of the liquid
and changes in pipe cross-section due to pressure change arising from
variations in pipeline elevation and, where resistance is included,
changes of piezometric level. This effect exists in steady flow as well
as under transient conditions and most investigators would normally
ignore this term although its significance is more important when hand-
ling modern plastic materials which are more prone to deformation.
4.4 Pipeline resistance Ð

The pipeline resistance term g S dt requires use of a resistance law
derived from steady flow measurements and it is assumed that this is
applicable under transient flow circumstances. For determining resistance
in a pipeline or tunnel system it is best to have on-site measurements but
this is not feasible in many instances, especially if it is a new scheme yet to
be constructed. Selection of an appropriate coefficient or coefficients
should err on the side of caution. In steady flow concerning say a
pumping main, typically for pump selection a roughness coefficient
would be chosen towards the upper limit of the range of likely values.
This ensures that the pump selected is able to develop the necessary
45
head. However, in transient studies where velocity changes are con-

cerned, pressure transients may be more severe if velocity is greater,
implying that roughness should be selected at the lower end of the scale
of likely values. Generally it is wise to study surging using alternative
roughness values to ensure that worst-case conditions have been
included.
Early experiments around 1850 on flow of water indicated that head
loss hf varied approximately directly with velocity head V2 =ð2gÞ and
pipe length L and inversely with pipe diameter D. Using a coefficient
of proportionality f, called the friction factor, Darcy, Weisbach and
others proposed an equation having the form:
hf ¼ fL=DV2 =ð2gÞ ð4:9Þ
This equation is commonly called the Darcy equation.
The value f was found generally to be dependant upon the relative
roughness k=D and on the Reynolds number Re ¼ VD= where is
the kinematic viscosity of the liquid ¼ dynamic viscosity/density.
Reynold’s experiments showed that viscosity caused two distinct flow
regimes to occur. At values of Re < 2100 flow was always laminar
while for Re > 4000 flow was always turbulent. For laminar flow all
turbulence will be damped by viscosity and no mixing will occur, while
for turbulent flow a chaotic motion of small fluid masses occurs in all
directions as flow takes place.
Under laminar motion the friction factor is given by:
f ¼ 64=Re ð4:10Þ
Probably the most accurate formula available for calculation of
pipeline friction factor f in the transition region between laminar and
fully developed turbulent flow is the Colebrooke—White equation as
described by Ackers (1963). Its implicit formulation makes it incon-
venient for application, that is:
p p
1= f ¼ 2 log10 ½2:51=ðRe fÞ þ k=ð3:71DÞ ð4:11Þ
where k is the equivalent uniform sand grain roughness height and
log10 represents the common logarithmic function. More convenient
approximations to the Colebrooke—White equation have been devel-
oped, for example:
f ¼ 0:25=½log10 f5:74=Re0:9 þ k=ð3:71DÞg2 ð4:12Þ
Alternative equations can be used, such as the Moody formula:
p
f ¼ 0:0055½1:0 þ 3 ð20 000:0k=D þ 106 =ReÞ ð4:13Þ
46
These equations allow f to be found with sufficient accuracy for analysis

purposes. Values of k for different pipe materials and conditions can be
found from sources such as Miller (1978).
The value of roughness height k may vary over time within a pipe-
line and consideration must be given to this aging process. Two
principal factors should be taken into account when assessing varia-
tions in k.
4.4.1 Corrosion
Metallic pipes such as steel and cast iron are liable to corrode. Hydraulic
Research Station paper No. 4 gives some data from studies by Cole-
brooke and White (Colebrooke, 1939) whereby the increase in k may
be estimated for older asphalted cast-iron pipes. The rate of increase
in roughness height with time represented by can be obtained from:
2 log10 ð12Þ ¼ 3:8 pH ð4:14Þ
where ¼ ft=year for cast iron asphalted pipes. The current value of
k is then found from the equation,
k ¼ ko þ t ð4:15Þ
t being pipe age in years and ko is the new pipe roughness height.
4.4.2 Sliming
Wastewater mains are liable to sliming. Recent research has shown that
k is primarily a function of velocity:
k ¼ V ð2:34Þ ð4:16Þ
where
¼ 0:054 (lower bound)
¼ 0:446 (mean) and
¼ 3:660 (upper bound)
Ninety-five per cent of observations lie between these upper and lower
limits.
From equation (4.16) it will be noted that as V ! 0 then k ! 0. To
avoid unreasonably small values of k it is recommended that a minimum
value of k be used equal to the value ko for the clean pipe.
For all types of wastewater rising mains, the following has also been
proposed as in the Biwater Manual (1988).
47
For V ¼ 1:0 m=s . . . 0:15 k 0:6

For V ¼ 1:5 m=s . . . 0:06 k 0:3
For V ¼ 2:0 m=s . . . 0:03 k 0:15
For sludges and suspensions, the friction factor at low flows, V < 1:2 m/s,
may be amended from the value for clean water.
4.4.3 Evaluation of the integral

Since Reynolds number is a function of velocity, it follows that f will
require to be recalculated as an analysis progresses.
Having found f then application of the Darcy equation (4.9) allows
resistance gradient S ¼ hf=L to be established, thus:
S ¼ f=DV2 =ð2gÞ
To preserve the correct sign of resistance gradient in a regime where
flow may reverse then:
S ¼ f=DVjVj=ð2gÞ
Ð
When evaluating the integral g S dt over some path length, a choice
has to be made regarding how velocity is to be represented:
ð ð
g S dt ¼ f=ð2DÞ V 2 dt f=ð2DÞVm 2
ðt1 t0 Þ or
f=ð2=DÞðV 2 Þm ðt1 t0 Þ
where ðV 2 Þm may be represented as 12 ðV12 p
þ V02 Þ; Vm may be represented
1
as 2 ðV1 þ V0 Þ the arithmetic mean, or as ðV1 V0 Þ the geometric mean,
Smith (1969). p 2
The representation
p of Vm as ðV1 V0 Þ has some advantage as Vm
becomes ½ ðV1 V0 Þ2 or simply V1 V0 . Preserving the sign of resistance
gradient then, this can be written V1 jV0 j. In the equations for the
quasi-invariant, ignoring the second small integral for the effect of
changing pipe elevation:
V g=aH ¼ J f=ð2DÞVjV0 jðt t0 Þ
or rearranging,
f1 ðÞf=ð2DÞjV0 jðt t0 ÞgV g=aH ¼ J ð4:17Þ
This formulation also preserves the linear nature of the equations.
48
5
Application of characteristic
equations
In the present chapter the equations derived in Chapter 4 are used in

the development of a practical method of computation of hydraulic
transient behaviour. Equations (4.2) and (4.4), obtained in the previous
chapter, describe the gradients of paths in the x—t plane along which
disturbances may propagate. In present circumstances the disturbances
are pressure waves propagating along the paths dx=dt ¼ V a with
quasi-invariant values J.
‘Characteristics’ of a system relate to a capacity for propagation rather
than to any specific propagation event. Thus a set of characteristic
paths can be deemed to exist although there may be no transient
event taking place. In other words these propagation paths will still
be present under both steady flow conditions and static conditions.
5.1 Use of the characteristics

If a wavefront is considered to be made up of an infinite number of
infinite wavelets or step changes in velocity and head, each of
infinitesimal magnitude (Fig. 5.1), then each wavelet will travel along
a characteristic. Assume the horizontal or x-axis is used to represent
distance along a pipeline and the vertical t-axis defines time elapsed
during a transient event. An infinite number of paths will exist of the
form Cþ or dx=dt ¼ V þ a within this x—t plane. Similarly there will
be an infinite number of C paths with gradient dx=dt ¼ V a
along which waves may travel in the opposite direction (Fig. 5.2).
The meeting of Cþ and C characteristics produces 12 intersection
points.
At any intersection of a pair of opposing characteristics, the invariant
relationships along these propagation paths yield two simultaneous
49
dx/dt = V ± a
Wavelets
Velocity of flow = V
Piezometric line – wavefront
Horizontal datum
Fig. 5.1. Representation of a pressure wave front
equations in V and H, that is:

V þ g=aH ¼ Jþ and V g=aH ¼ J
giving
V ¼ ð Jþ þ JÞ=2 ð5:1Þ
and
H ¼ ð Jþ JÞ=ð2g=aÞ ð5:2Þ
Consider any typical intersection point P in this x—t plane. If a
transient event is initiated at point P, the effects will be propagated
along the Cþ and C paths which pass through the point and an
increasing length of pipeline will be affected by the event with the
passage of time. The wedge-shaped area bounded by these Cþ and
C paths is called the region of influence of P, Abbott (1966, p. 17),
and is represented by the long-dashed lines in Fig. 5.2. Flow conditions
at points outwith this area are not changed by the event at P.
Also if earlier transient conditions are travelling along the Cþ and
C characteristics which meet at P then the resulting flow conditions
50
Application of characteristic equations
C– characteristics
dx/dt = V – a
C+ characteristics
dx/dt = V + a
+t
Limits of region
influenced by P
Typical intersections
Limit of region upon

which P depends
+x
Fig. 5.2. Characteristics in the x—t plane
at this point will be determined only by conditions within the area

bounded by the characteristics represented by the short-dashed lines
of Fig. 5.2. This region is sometimes called the domain of dependence
of P. Points outwith this area cannot influence conditions at P.
The characteristics and their quasi-invariant relationships can be
used to provide a solution to the task of predicting propagation of pres-
sure waves. For practical purposes a representative set of characteristics,
defined by the heavier lines in Fig. 5.3, is chosen from the 1 number
available in each direction. A sufficient number of paths are required
to allow the pressure waves to be modelled with accuracy.
Solutions can be found at each point where opposing Cþ and C
characteristics intersect. Values of quasi-invariants Jþ and J are
established from earlier locations on each characteristic. The process
of solution starts from some set of known flow conditions thoughout
the network. These ‘initial’ conditions may be either a static or
steady flow state or even transient flow conditions which may have
been determined from a previous simulation.
51
Representative characteristics
C– C+
+t
+x
Fig. 5.3. Representative characteristics in the x—t plane
A number of options exists as far as choosing the representative set of

characteristics. Several of these will be described with the advantages
and disadvantages of each mentioned.
5.2 ‘Natural’ characteristic mesh

An initial set of paths is chosen and the progress of these followed as
time passes (Fig. 5.4). The intersection points of opposing paths
provide solutions for transient flow conditions. Values obtained are
accurate and this approach has the capability of detecting shock
fronts and to follow their progress. While a regular distribution of
characteristics may be chosen at the outset, as time passes the distribu-
tion of intersection points at which solutions are found becomes
increasingly irregular, particularly where appreciable variations in
wave speed occur as when gas release takes place. For a simple pipeline
having uniform properties along its length, this method poses no
52
C– paths
C+ paths
+t
+x
Fig. 5.4. ‘Natural’ or irregular characteristic mesh
significant difficulties and has been used effectively by Larsen (1976) for
example, in his investigation of undersea outfalls. Considerable
complexity is introduced where a looped or branching system is
encountered or where internal changes of pipe material or diameter
are present for instance. Although not insurmountable, the computa-
tional difficulties of following progress of characteristics through more
complicated systems are sufficient to make this approach unattractive
for general use.
The irregular distribution of intersection points at which solutions
are obtained also means that some post-processing is required to
make the predictions easy to understand. Typically the end-user
wishes to see time variations of pressure, head, flow, etc., at specified
locations and this requires interpolation within the irregular mesh of
solutions.
Almost all computational schemes designed for general use employ a
regular or ‘fixed’ mesh arrangement which provides predictions at
predefined locations throughout the pipeline system and at regular
increments of time. Characteristic paths are selected, from the 1
53
available in each direction, which intersect at the predefined mesh

points. There remain considerable differences between alternative
fixed-mesh schemes with much of the difference centred on how the
wave speed is handled.
5.3 Using variable wave speed a

The equation for acoustic velocity or wave speed a has been given in
Chapter 3 as:
p
a ¼ 1= fm ðDc1 =ðsEÞ þ ð1 Þ=K þ =ðpabs Þg ð3:7Þ
where m ¼ ð1 Þ þ g .
The wavespeed given by equation (3.7) is a composite involving
three components: the solid pipe or tunnel wall, the liquid component
and the gaseous component. Variability of the wave speed with chan-
ging pressure for a given gas content produces changes in characteristic
gradients dx=dt ¼ V a.
Use of a fixed grid as shown in Fig. 5.5 allows computation to be
carried out in a way that yields values of the dependent variables flow
rate, pressure and gas content at regular intervals along each pipeline
of a system and at fixed time intervals. As the wave speed changes
with pressure during a transient event, especially at low pressure,
then the gradients:
dx=dt ¼ V a
will alter and it is necessary to interpolate on the previous time line to
establish values for the Riemann quasi-invariants. The solution can
then be obtained at the intersection points of opposing pairs of charac-
teristics. Most schemes rely on the interpolation points being within the
pipeline sections directly on each side of the solution point. This
requires that the distance increment x and the time increment t
be chosen so that the quotient:
x=t jV ajmax ð5:3Þ
This Courant Lewy stability criterion ensures that interpolation takes
place between the adjacent grid points. Any appreciable and continued
extrapolation beyond these grid points is likely to lead to instability. The
interpolation point should lie as close to the grid point as possible but
without falling outside the reach of the pipeline. Fox (1977) describes
use of such a scheme and, based upon experience, has found that
x=t ¼ a=0:95 works well. The process of interpolation leads to
54
Dt
+t C– characteristics C+ characteristics
dx/dt = V – a dx/dt = V + a
Dt
Dx Dx
+x
Fig. 5.5. Fixed-grid characteristic mesh
some dispersion of the wave front and there is an element of compro-

mise involved between minimising the effects of numerical dispersion
while retaining the ability to model variable wave speed.
5.4 Use of a larger time step

A numerical approach which can permit smaller quotients or effectively
larger time steps t for the same x increments:
x=t < jV aj ð5:4Þ
was described by Vardy (1976). This involves a more complicated
routine to identify the position of the interpolation point as shown in
Fig. 5.6. While not restricted to variable wave speed schemes, this
approach can retain the ability to accommodate variable speed of
55
Dt
+t
C+ characteristics
Dt
Dx Dx
+x
Fig. 5.6. Fixed-grid mesh with extended time steps
propagation of the pressure waves but contains the same risks of disper-
sion of the wave front.
5.5 Use of a fixed wave speed

Streeter and Wylie (1967) used a different approach. The effect of
velocity upon the characteristic gradient dx=dt was neglected, and a
was taken as constant, giving:
dx=dt ¼ a ð5:5Þ
By setting:
x=t ¼ dx=dt ¼ constant
56
Cross-sectional area = A Gas volume = aADx
Vu/s Vd/s
Dx/2 Dx/2
Dx Dx
+x
Final volume = (a1 + af)ADxhatm/habs
Vu/s Vd/s
C+ characteristic C– characteristic
Dt
Vu/s(o) Vd/s(o)
Initial volume = a1ADxhatm/habs(o)
Fig. 5.7. Representation of free gas in fixed-grid computations
the need for interpolation is avoided and with it the effects of any
numerical dispersion.
In this scheme, free air or gas released from solution, which can influ-
ence the wavespeed a, is handled by ‘collecting’ all of the gas in the
neighbourhood of a computing point, into a single ‘bubble’ occupying
the cross-section of the pipe at a local computing point (Fig. 5.7). A
discontinuity of velocity (and flow) occurs across the bubble which
expands and contracts as pressure changes. Behaviour of the gas mass
is represented using a gas law.
If the gas void fraction in the neighbourhood of a solution point and
at the start of a time increment t is o at absolute pressure pabsðoÞ and
57
it is assumed that this fraction is representative of 12 x on each side of

the solution point then the total gas volume at the solution point will be
0 xA cubic metres.
The following equations are available to provide values of upstream
and downstream velocities Vu=s and Vd=s , piezometric level H and
final void fraction at the end of the time increment.
From the advancing Cþ characteristic, Jþ ¼ Vu=s þ g=aH and from
the backward C characteristic, J ¼ Vd=s g=aH from conservation
of volume across the bubble over the time increment,
fVd=s þ Vd=sðoÞ ðVu=s þ Vu=sðoÞ ÞgA=2 ¼ A x d=dt ð5:6Þ
where void fraction is given by:

¼ ð1 þ f Þhatm =habs
d=dt ¼ hatm dð1 =habs Þ=dt þ hatm dðf =habs Þ=dt
if free gas content is o at the start of the increment then,
hatm dð1 =habs Þ=dt hatm o ð1=h2abs ÞðH Ho Þ=t
and
hatm dðf =habs Þ=dt hatm =habs df =dt
where the derivative df =dt is the rate at which gas is evolving from
solution. This is a quantity that can range from all available gas being
freed as used by Fox (1977) to a more controlled rate of release. The
relationship between piezometric level or total head H and the absolute
pressure head habs is:
habs ¼ H z þ hatm
These equations are sufficient to provide solutions for the flow variables
at the close of the time increment.
This approach avoids the need for interpolation and thus any risk of
numerical dispersion. However, even where interpolation is used, this is
mainly significant only at low pressures when gas bubbles are forming,
with pressure changes being modest so that any errors introduced to
interpolation are not likely to be very significant.
5.6 Distribution of free gas along the pipeline

Other implementations of the equations have also used the fixed
regular grid but handled the matter of wave speed and gas release in
different ways. For example, a constant wave speed is maintained in
58
each pipe, based upon the composite pipe and liquid system. Gas release
is modelled as a separate phase which may be either distributed along
the top of the pipe in the form of a continuous gaseous phase, or
could be represented as a discrete bubble where it occurs at a well-
defined summit on the pipeline. A flexible model will include the
option of representing free gas or vapour in one or both of these forms.
5.7 Model output

Nowadays, rarely does the user require to consider the detailed schema-
tisation of a pipeline or the relationship between x and t. These
aspects will often be resolved by the computer program itself.
Using one of the approaches described in sections 5.3 to 5.6, values of
head H, pressure ¼ gðH zÞ, velocity V and flow rate can be
predicted at each of the intersection points of the characteristics. For
each simulation this process is continued until all significant effects
have been modelled. It is then a relatively straightforward matter to
present these predicted values in forms which are readily understood.
Both tabular and graphical presentations are considered important.
Plotted output usually can take two forms. First, time histories of
specific variables such as flow, pressure or air volume at defined loca-
tions within a network for instance, allow timescales of events to be
easily identified and the changes at one place to be related to
changes at other points. Amplitude of pressure transient oscillations
can also be seen in these plots. The second form of graphical presenta-
tion is ‘envelope’ curves showing the variation of extremes of head or
pressure, for example, throughout a pipeline. This type of plot allows
the engineer to identify zones of sub-atmospheric pressure or regions
where peak pressure has exceeded allowable limits.
Tabulated information on extremes of pressure and flow avoids the
need to scale information from graphical output and complements
the plotted forms of presentation.
59
6
Boundaries
In earlier chapters it was described how the task of determining flow and
pressure changes along a pipeline could be achieved by using the
network of characteristics and the values propagating along these
paths. The challenge remains of finding corresponding values of flow
parameters at the many and various features and fittings which are to
be found in most pipeline networks. Each feature will act to transmit
and/or reflect transient pressure waves in a manner which is particular
to that type of fitting or element of the pipeline. Each of these elements
is referred to as a boundary. Analysis of response of these features
requires that an additional ‘auxiliary’ equation or equations has to be
introduced to complete the process of solution and this chapter intro-
duces just some of the wide variety of possible pipeline elements and
the corresponding equations.
At each computing point within a pipeline, values of velocity V and
total head H were obtained through solving a pair of simultaneous equa-
tions at the intersection point of characteristics travelling in opposing
directions within the pipeline. At the upstream and downstream extremi-
ties of a pipeline only a single characteristic intersects the boundary at any
instant of time, as shown in Fig. 6.1. Since values of both V and H are
required it is necessary to introduce additional equation(s), sometimes
referred to as auxiliary equations, to provide a complete description of
conditions at the extremity of the pipeline. The nature of boundaries is
varied and it is worthwhile spending a little time in considering some
of the types and common assumptions which are made regarding these.
As a rule the steady flow characteristics of a boundary such as a valve
or pump are assumed applicable to unsteady conditions.
Boundaries usually represent some actual physical feature of a
network but may also be numerical devices introduced for convenience
60
Boundaries
Boundary node Internal nodes
Characteristics Dt
Single characteristic Dt
at boundary
Dx Dx
Fig. 6.1. Characteristics at a boundary
or to simplify computations. Examples of both types are presented in

this chapter. While some boundaries occur at the physical extremities
of pipelines such as where a pipeline discharges to a tank or reservoir,
others may occur partway along a main. Such internal boundaries can
include in-line valves, booster pumps and self-acting air valves and
pressure relief valves.
6.1 Types of boundary

It is at boundaries that hydraulic transients can be introduced to the
pipeline network. A valve opening or closing represents one example
of a transient initiating boundary. One of the most common boundaries
which can introduce transient effects is that of a pump starting or
stopping. A pipe burst could be represented as a valve fitted to the
side of the pipe and this also can be used as a means of including
surge effects.
Where a pump remains in constant speed or where a valve setting
is unaltered then these forms of boundary act as devices which reflect
61
and transmit pressure waves to an extent dependent upon the charac-

teristics of the particular hydraulic element. As a general rule other
pipeline features such as tanks and reservoirs, blank flanges, changes
of pipe properties and bifurcations for instance all act passively to
reflect and/or transmit pressure waves. A rare exception to this rule
might be where surface waves introduce pressure and flow variations
at the inlet of a submerged pipe.
Other boundaries such as air-valves, pressure-relief valves and check
valves respond to changing flow conditions and may only come into play
at certain stages of a transient event. They will have little if any influ-
ence on propagation of pressure transients at some times while they may
be responsible for producing important secondary pressure surge events
at another stage of the hydraulic transient.
Additional boundaries are also required to represent pressure tran-
sient suppression equipment. Some of the equations used to model
response of such equipment are discussed when considering specific
forms of protection in later chapters.
A further type of boundary condition is that which is introduced
by the modeller for convenience to simplify or reduce the work of
analysis. This form of boundary is purely a numerical device and need
not correspond with any physical feature of the pipeline network.
A common assumption with most models is that at any instant
of time, flow behaviour at the boundary is identical to that which
would occur under steady flow. For example, the head drop through
a valve or other fitting at some transient flow rate Q ¼ QðtÞ would
be taken as identical to the head loss if Q were a constant flow
rate.
6.2 Reservoirs and tanks

Piezometric level Hr at a pipe inlet may be taken as constant and equal
to the free-surface water level as in the case of a large reservoir, or as
relatively slowly varying with time or possibly changing periodically
with surface wave action.
When flow is from the pipeline into the reservoir piezometric level
just inside the pipe, H will effectively be the same as in the reservoir,
thus H ¼ Hr .
Taking that characteristic arriving at the pipe inlet for time t
(Fig. 6.2), then the quasi-invariant relationship gives:
V g=aH ¼ J or V ¼ J ðÞg=aHr
62
Boundaries
Reservoir level = Hr
Pipeline
Dx
Boundary node
Internal node
Dt
Characteristics
Single characteristic
arrives at boundary
Fig. 6.2. ‘Large’ reservoir boundary
When flow is into the pipeline from the reservoir, kinetic energy is
developed and losses are incurred so that Hr ¼ H þ ð1 þ KL ÞV2 =ð2gÞ,
where KL is the entry loss coefficient. The relationships at the pipe
inlet at any time t become:
V g=aH ¼ J and H ¼ Hr ð1 þ KL ÞV 2 =ð2gÞ

Substituting for H then:
ð1 þ KL Þ=ð2gÞV 2 ðÞa=gV þ a=g J Hr ¼ 0 ð6:1Þ

This quadratic equation can be readily solved for V and, by substituting in
the quasi-invariant relationship, H can be determined. For further details
of the solution see section 6.5 below. In many instances of waterworks
practice, velocity is relatively low so that ð1 þ KL ÞV 2 =ð2gÞ is small. For
instance if V ¼ 1 m/s and KL ¼ 0:5, ð1 þ KL ÞV 2 =ð2gÞ ¼ 7:6 cm. At
relatively low velocities it is quite acceptable to set H ¼ Hr for both
63
inflow and outflow. Then the solution for velocity V is simply:

V ¼ J ðÞg=aHr ð6:2Þ
In other circumstances, for instance in hydropower applications or
when studying transients in the outflow pipelines of high dams, the
velocities can be greater by an order of magnitude or more. If velocity
is 30 m/s — not unusual for dam outlet works — V 2 =ð2gÞ ¼ 45:9 m,
which clearly cannot be neglected and the quadratic form of equation
(6.1) is necessary.
6.3 Branches and changes in pipe properties

Branches are a common feature of many pipeline networks. Consider a
simple connection of three pipes as illustrated in Fig. 6.3a.
Each of the three pipes has a boundary at the junction of the set of
pipelines. The x—t planes for each pipe can be shown as meeting
along a common time axis as depicted in Fig. 6.3b. In general the
positive direction of flow in any pipe may be directed either towards
the junction or away from the junction. Accordingly the Riemann
quasi-invarient travelling along a characteristic towards the junction
may be either of the Jþ or the J type depending upon the assigned
þve direction of flow in that pipe.
From the infinity of characteristics approaching the junction along
any pipe i, that path is chosen which arrives on the common time
axis at the instant t when a solution is to be found. The characteristic
for pipe i will provide a quasi-invarient value Ji at time t on the common
time axis. An equation of the form:
Vi g=ai H ¼ Ji
is thus obtained for each characteristic at the junction. A common
piezometric head H has been assumed at the junction end of each
pipe — that is, the velocity head in each pipe is considered to be
negligible. In water supply, economic velocities are normally of the
order of 1 m/s while sewage rising mains will usually carry flows at
higher velocities, say up to 3 m/s or more, often to ensure self-cleansing.
Velocity head will thus be of the order of 0.05—0.5 m. Ignoring the
change in velocity head from one pipe to another in these conditions
will in general not lead to significant error. Other applications may
require more careful consideration of velocity changes; for example, in
the pressure pipelines or penstocks of a hydroelectric plant, the flow
velocity can be much higher with velocities through relief valves as
64
Boundaries
Piezometric lines for pipes 1, 2 and 3
Common head = H measured

Area = A1 from horizontal datum
Area = A2
Pipe No. 1
Pipe No. 2
Pipe No. 3
Positive direction of flow may be
defined as towards the junction or
away from the junction in any pipe
Area = A3
(a)
Characteristics arriving at the junction
C2
Dt
C1
1 C3
Dx1
Dx2
Dx3
(b)
Fig. 6.3. Representation of a pipe junction
high as 60 m/s. Velocity at the emergency closing valves of a dam can be as

high as 30 m/s. Velocity head V 2 =ð2gÞ then becomes much more substan-
tial in these circumstances. Supposing a pipe junction is made between
pipes having cross-sectional areas A1 ¼ 1 m2 and A2 ¼ 2 m2 (Fig. 6.4).
At a flow rate of 20 m3 /s the change in piezometric level between the
65
Piezometric line or hydraulic gradient
Common head = H
Area = A2
Area = A1
V1 V2
Wave speed = a1
Wave speed = a2
+x
Dt
Dx1 Dx2
Fig. 6.4. Closed-end or blank flange boundary
pipes, ignoring losses, is 15.3 m — a quantity which cannot be ignored.

Consideration of energy losses and of kinetic energy changes becomes
necessary and calculation of piezometric level changes based on the
assumption of common head at the junction is no longer acceptable.
Returning to the simpler case where it is acceptable to neglect
velocity head changes at the junction then ignoring compressibility of
the relatively small volume of water at the junction, conservation of
volume yields:
X
ðAi Vi Þ ¼ 0 ð6:3Þ
where the summation is over all contributing pipes. Substituting for Vi
then:
Vi ¼ Ji ðÞg=ai H
66
Boundaries
and in the conservation equation:

X
½Ai ð Ji ðÞg=ai HÞ ¼ 0
or
X X
ðAi Ji Þ ¼ g ðAi =ai ÞH
giving,
X X
H ¼ 1=g ðAi Ji Þ= ðAi =ai Þ ð6:4Þ
If the assigned positive direction of flow in pipe i is towards the junction
then the plus sign is used for that pipe otherwise the minus sign is used.
Through induction, equation (6.4) can be shown to apply to any
number of pipes m, where 1 m 1, meeting at a junction. This
equation (6.4) for the junction of m pipes is the simplest form obtain-
able and this equation can be enhanced to include the presence of
air or gas at the junction, or isolating valves on each branch, which
may be either open or shut.
The corresponding velocity at the junction end of any pipe i at
time t is given by substituting the value of H obtained above, into
the equation for the quasi-invarient value Ji at the junction end of
the pipe. Thus:
Vi ¼ Ji ðÞg=ai H ð6:5Þ
6.3.1 Specific cases — number of pipes ¼ 1

When the number of pipes ¼ 1 the equation represents the solution at a
closed end or blank flange (Fig. 6.5). Thus:
H1 ¼ 1=gðÞA1 J1 =ðA1 =a1 Þ or H1 ¼ a1 =g J1 ð6:6Þ
and
V1 ¼ J1 g=a1 H1 or V1 ¼ J1 g=a1 a1 =g J1 or V1 ¼ 0 ð6:7Þ
6.3.2 Specific cases — change of cross-sectional area

When the number of pipes ¼ 2 the equation can be used to represent
some change of pipe properties, either a change of cross-section,
material, wall thickness or resistance. With reference to Fig. 6.5, then
ignoring changes of kinetic energy and losses at the junction of pipes:
H ¼ 1=gðþA1 J1 A2 J2 Þ=ðA1 =a1 þ A2 =a2 Þ ð6:8Þ
67
Pipeline Blank flange or closed end
Internal node
Boundary node
Dt
Single characteristic
at boundary
Dx
Fig. 6.5. A change of cross-section
Velocities in pipes 1 and 2 are respectively given by:

V1 ¼ J1 g=a1 H and V2 ¼ J2 þ g=a2 H ð6:9Þ
Any change of pipe material, wall thickness or diameter will lead to a
change of wave speed a. Wave speed may also be altered by the
pressure-dependent gas concentration in a pipe.
A special case of a change in cross-sectional property is where a pipe-
line terminates at a reservoir (Fig. 6.2). Suppose the þve direction of
flow is into the reservoir. Within the reservoir V2 ¼ 0 ¼ constant
and H2 ¼ Hr ðtÞ where Hr ðtÞ is a slowly varying function of time or
may be constant. The reservoir cross-section A2 is very large com-
pared to pipe area A1 and wave speed in the reservoir can p be taken
as the speed in an unconfined body of water — that is, a2 ¼ ðK=Þ,
then:
J2 ¼ V2 g=a2 H2 ¼ g=a2 Hr
H ¼ 1=gðþA1 J1 A2 J2 Þ=ðA1 =a1 þ A2 =a2 Þ
H ¼ 1=gðþA1 J1 A2 g=a2 Hr Þ=ðA1 =a1 þ A2 =a2 Þ
68
Boundaries
since
A1 =a1 A2 =a2 and jA1 C1 j jA2 g=a2 Hr j
H ¼ 1=gðA2 g=a2 Hr Þ=ðA2 =a2 Þ
or
H ¼ Hr
6.4 Response of a large pipe or trunk main

It is not uncommon for work to be carried out on parts of an extensive
network of interconnected pipelines. Pressure transients are no
respecters of contract boundaries and will spread throughout a system
of pressure pipes until a shut valve/closed end or a reservoir is
encountered. From each of these the pressure wave may be taken as
reflected 100% in one form or another. If the entire network were to
be modelled then the workload could be very considerable, not least
in the effort required to obtain adequate information on other parts
of a large system outwith the contract area of immediate interest. It
may be acceptable, in certain circumstances, to limit the extent of
modelling by an approximation at some important point of wave reflec-
tion within the system as a whole.
Suppose, for example, a large pipeline, trunk main or aqueduct supplies
a number of smaller branch networks as illustrated in Fig. 6.6. Suppose
work is to be carried out on one of these branched systems including a
hydraulic transient investigation. Pressure transients created within the
branched system will propagate upstream towards the connection of
the branch with the main aqueduct. Assume an initial steady velocity
Vo exists in the branch main just downstream of the connection. If the
initial steady-state piezometric level at the junction is Ho then setting
the datum for head measurements at this level Ho ¼ 0. Should the
initial velocity be reduced to zero by valve closure in the branch
network then the inertial head rise Hb in the branch main will be:
Hb ¼ aVo =g ð6:10Þ
Ignoring changes due to resistance and pipe elevation variation, then
the invariant value J leaving the shut valve just after its closure, is
given by:
J ¼ V g=aH ¼ 0 g=aða=gVo Þ ¼ Vo ð6:11Þ
In the upstream section of the aqueduct, assuming all flow entering
the branch comes from this direction and that no other flows are
69
Aa
+x
Ab
+x
+x
J+ = VoAb/Aa
J– = 0
J– = –Vo
Fig. 6.6. Small branches connected to trunk main
present:
Jþ ¼ Vo Ab =Aa þ g=a0 ¼ Vo Ab =Aa ð6:12Þ
And from the downstream section of aqueduct where zero flow is taking
place:
J ¼ 0 þ g=a0 ¼ 0 ð6:13Þ
Due to valve closure in the branch main, the head change Hj at the
junction of the branch main with the aqueduct is given by equation
(6.4). Substituting in this equation then:
Hj ¼ 1=g½ðVo ÞAb þ Vo Ab =Aa Aa 0Aa =ðAb =ab þ Aa =aa þ Aa =aa Þ
or
Hj ¼ 2=gAb Vo =ðAb =ab þ 2Aa =aa Þ ð6:14Þ
Head rise at the junction as a proportion of the initial head rise at the
valve in the branch is then:
Hj =Hb ¼ 2=½1 þ 2ðAa =aa Þ=ðAb =ab Þ
70
Boundaries
or for
a ¼ constant Hj =Hb ¼ 2=ð1 þ 2Aa =Ab Þ
Inserting values for area and wave speed into the above equation it
can be shown that as aqueduct area/branch area increases the response
at the connection tends more and more towards that of a simple reser-
voir. This relationship allows the investigator to assess the implication
of replacing the aqueduct connection by a simpler boundary such as a
reservoir.
6.5 Actuated valves and pipeline fittings

In the case where isolating valves such as gate valves or butterfly valves
are distributed along a pipeline at intervals to facilitate maintenance
and these valves remain open throughout a transient event, the head
drop through each valve is relatively small and so there is little to be
gained by modelling each valve as an internal boundary. It is usually
sufficient to establish the overall head loss coefficient for all of the
valves and to distribute this along the pipeline together with other
minor losses from bends, tees, etc., to yield an overall resistance
coefficient. Overall head loss Hf is given by:
nX o
Hf ¼ ðKL Þ þ fL=D V 2 =ð2gÞ ¼ fe L=DV 2 =ð2gÞ ð6:15Þ
P
where fe ¼ f þ KL D=L.
Other valves or fittings may impose significant head losses on the
flow even when fully opened and so warrant inclusion in a model as a
discrete head loss. Where a valve may operate to open or close
during a transient event it is essential to include this as a discrete
head drop in the model.
In general a manufacturer will provide data on the relationship
between flow rate and head loss through a valve in the form of a
head loss coefficient curve as a function of valve position. The head
loss coefficient KL is documented as a function of valve stroke between
the fully open position and some point near to the shut position.
Figure 6.7 depicts typical head loss coefficients for a range of valve
types. Due to the very non-linear relationship between KL and stroke,
particularly as a valve nears the closed position, it is not advisable to
interpolate values of KL directly. Rather the head loss relationship
should be represented in a way that endeavours to reduce the p degree
of non-linearity. A simple approach is to interpolate values of 1= KL .
71
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0
1000 n
Valve type* D (in.) ao/A Port shape
A Gate 12 1.0 Circular
B Butterfly 24 0.65 –
C Globe 6 1.0 Shaped (iii)
D1 'In-line' regulator 12 1.0 Rectangular
D2 'In-line' regulator 9 Shaped (i)
100 i ii n iii
n
n
Port shapes
* See Fig. 4
Valve headloss coefficient, Kn
D2 † Model
E
Valve full open
C
B F
10 10
D1
G
A
Ko
1 1
Valve type* D (in.) ao/A Port shape

E Submerged disch.
regulator 6 1.0 Rectangular
F Sleeve regulator 12 1.0 Shaped (ii)
G Free disch. regulator 8* 2.0 Rectangular
0.1 0.1
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0
Valve position as a proportion of port stroke, n
(a)
D In-line regulator G Free discharge regulator
E Submerged discharge regulator F Sleeve regulator

(b)
Fig. 6.7. (a) Head-loss coefficients for different valve types; (b) control valve type
72
Boundaries
6.5.1 Terminal valves

Consider a valve located at the extremity of a pipeline as shown in
Fig. 6.8. At any instant of time a characteristic will arrive at the
valve from the pipeline providing the quasi-invariant relationship
between H and V at the valve:
V g=aH ¼ J
A second relationship is provided by the head loss equation for the
valve:
H ¼ Sj Sv KL V 2 =ð2gÞ ð6:16Þ
where H ¼ H Hr , Sj ¼ þ1 for a Cþ characteristic and Sj ¼ 1 for
a C characteristic; also Sv ¼ Vo =jVo j. Substituting for Hr then:
Sj ½Sj Sv KL =ð2gÞV 2 þ Hr g=a þ V J ¼ 0
or rearranging,
Sv KL =ð2aÞV2 þ V þ Sj g=aHr J ¼ 0
Solving for V then:
p
V ¼ Sv a=KL ½ f1 þ 2Sv KL =að J Sj g=aHr Þg 1 ð6:17Þ
Discharge level = Hr
Piezometric level = H
Pipeline Valve
Internal node
Boundary
Dt Single characteristic
at boundary
Dx
Fig. 6.8. Valve at a reservoir or tank
73
6.5.2 In-line valve

Upstream and downstream of the valve are stretches of pipeline within
which pressure transients can propagate (Fig. 6.9).
From the upstream and downstream pipelines characteristics will
arrive at the valve at any time t. The invariant relationships are:
Vu=s þ g=a1 Hu=s ¼ Jþ for the Cþ characteristic

and Vd=s g=a2 Hd=s ¼ J for the C path
Across the valve:
Hd=s Hu=s ¼ Sj Sv KL =ð2gÞV 2
and from conservation of volume:

Vu=s Au=s ¼ VA ¼ Vd=s Ad=s
Piezometric line
Hu/s
Valve Hd/s
Dt
Dx Dx
Fig. 6.9. In-line valve
74
Boundaries
Eliminating Hu=s , Hd=s , Vu=s and Vd=s then writing:

constant ¼ Sv A=KL ðau=s =Au=s þ ad=s =Ad=s Þ
p
V ¼ constantð1 f1 2Sv =KL constantðau=s Jþ þ ad=s JÞg
ð6:18Þ
For both the terminal valve and the in-line valve additional measures
have to be taken to accommodate the gaseous void fraction and possible
vapour cavity formation.
6.5.3 Automatic control valves

Valves may be installed for the purpose of regulating pressure and/or
flow within part of a network. Various alternatives are possible and
some of these are considered.
6.5.3.1 Pressure-reducing valve

This valve is set to reduce a constant or variable inlet pressure and to
provide a constant outlet pressure. The valve is adjusted to produce a
constant outlet pressure head Hc or piezometric level so that with
reference to Fig. 6.9, Hd=s ¼ Hc . Then the C characteristic yields
an invariant value,
J ¼ V g=aHc or V ¼ J þ g=aHc
Then on the inlet side of the valve the Cþ characteristic gives,
Jþ ¼ V þ g=aHu=s
or substituting the value of velocity,
Hu=s ¼ a=gð Jþ VÞ ð6:19Þ
When the valve is fully opened,
Hd=s ¼ Hu=s Ko V 2 =ð2gÞ ð6:20Þ
where Ko is the valve head loss coefficient when fully opened. If
Hd=s < Hc , control cannot be maintained and a solution for the fully
opened valve is used.
6.5.3.2 Pressure-sustaining valve

This valve is set to maintain a constant inlet or upstream pressure so
that Hu=s ¼ Hc . The Cþ characteristic yields a quasi-invariant value
75
Jþ at the valve inlet so that,

Jþ ¼ V þ g=aHc or velocity V ¼ Jþ g=aHc
Then on the outlet side of the valve,
J ¼ V g=aHd=s or Hd=s ¼ a=gð J VÞ ð6:21Þ
When the valve is fully opened,
Hu=s ¼ Hd=s þ Ko V2 =ð2gÞ ð6:22Þ
If Hu=s > Hc then the downstream head is too high to allow the set head
to be maintained. If the valve is shut then V ¼ 0:0 and at the valve
inlet,
Jþ ¼ 0 þ g=aHu=s or Hu=s ¼ a=g Jþ
If Hu=s < Hc then the downstream head is too low to allow the set inlet
head to be maintained.
6.5.3.3 Demand-sensing pressure-reducing valve

In this configuration a minimum or base outlet pressure or deliver head
Hb is set. This head will apply at zero flow. The valve is adjusted to
produce a delivery head variation of the form,
Hd=s ¼ Hb þ KL V2 =ð2gÞ
where KL is a coefficient defining the relationship between velocity and
head. Then on the delivery side of the valve,
J ¼ V g=aHd=s ¼ V g=a½Hb þ KL V 2 =ð2gÞ
Rearranging,
KL =ð2aÞV 2 V þ g=aHb J ¼ 0
or
p
V ¼ a=KL f1 ½1 2KL =aðg=aHb JÞg ð6:23Þ
then
Hd=s ¼ Hb þ KL V2 =ð2gÞ
On the inlet side of the valve,
Jþ ¼ V þ g=aHu=s or Hu=s ¼ a=gð Jþ VÞ
76
Boundaries
For a viable result Hu=s Hd=s Ko V2 =ð2gÞ where Ko is the fully open
valve head loss coefficient.
6.6 Use of more than one time step

If the modelling exercise is only one part of an overall study of hydraulic
transients in a system then the computer model may reasonably be
expected to encompass the entire network. As pipe lengths within a
pumping station, for example, may be quite short, this implies a small
time increment which may not be particularly suitable for the much
larger pipe lengths of the majority of the network. A modified com-
putational scheme which was described by Vardy (1976) may be
useful in this context. The current author has used this technique on
a number of occasions in these circumstances. Consider Fig. 6.10.
Two distinct time steps are in use. A small time increment is
employed within the vicinity of the pumping station so that
xp =tp jV aj for the small lengths of pipe x. Outwith the
Pipeline
Dtp
Dtp
Dt
Dtp
Dtp
Dtp
Dxp Dxp Dxp Dxp Dx
Fig. 6.10. Inclusion of a change in time step
77
pumping station, larger x increments can be used with correspond-

ingly increased t.
Computations in the pumping station are carried out using the fine
grid spacing xp and the corresponding small time step tp , while in
the system as a whole a larger increment size x and time increment
t are used. The relationship between the time increments is:
t ¼ m tp
where m is a positive integer.
The ‘boundary’ between use of large and small increment sizes is
placed at a location such that any significant factors which could
have an influence on transient behaviour in the pumping station are
included in the fine grid part of the system. These factors might
include pressure vessels and air valves for example. If a booster
pumping station with significant lengths of suction main is being
modelled, a second upstream ‘boundary’ or interface is required
between the main grid spacing and the fine grid spacing.
Transients developed within the pumping station are modelled in
greater detail than are events in the overall system. Once a pressure
wave has passed through the interface into the system as a whole,
detail will be lost where wave components have a short period <t.
This will not materially influence predictions of events within the
detailed area at least for some time until wave reflections from
outwith the fine grid area return to the ‘boundary’.
6.7 Non-reflecting boundary

If interest is centred on transient behaviour initiated within the detailed
area alone and transient behaviour in the system as a whole does not
require to be modelled then it may be possible to terminate the computer
model at a section local to the area of interest, say a pumping station.
This can be done when it is considered that the transient event of
interest occurs in a time t the wave reflection time from the nearest
feature which will produce a significant response (Fig. 6.11) — that is,
time of modelling t 2L=a. In this instance a non-reflecting boundary
can be introduced close to the pumping station. Suppose we have a
boundary to which no pressure wave effects arrive from one side and
the Riemann invariant remains at steady-state values, for example
steady design flow rate. Figure 6.12 illustrates this circumstance.
Let the initial conditions at a non-reflecting boundary be represented
by velocity Vo and piezometric level Ho . Then the invariant value is
78
Boundaries
Part of pipeline to be modelled

First point of significant reflection
Reflection time = 2L/a
Non-reflecting boundary
Fig. 6.11. Simple non-reflecting boundary
Dx
Jp = Vp ± g/aHp
Characteristics without
the area being modelled
Dt
Jo = Vo ± g/aHo
Jo = Vo ± g/aHo
Jo = Vo ± g/aHo
Characteristics within the
area being modelled
Initial conditions Vo and Ho
Fig. 6.12. Characteristics at a non-reflecting boundary
79
given by:
Jo ¼ Vo g=aHo ð6:24Þ
If no reflection or transient effects arrive at the boundary from the main
network then the value Jo will remain the same regardless of the time
which has elapsed. From within the modelling area the pressure wave
effect from the transient event will be given by the prevailing values
of velocity Vp and Hp as shown in Fig. 6.12. So that:
Jp ¼ Vp ðÞg=aHp ð6:25Þ
is the quasi-invariant value at the end of the segment of characteristic
leading to the point t at which a solution is to be found. The effects of
friction and any other influences along the segment of path have been
added to yield this final value of Jp . The solution after any time step is
then given by:
V ¼ ð Jp þ Jo Þ=2 and H ¼ ð Jp Jo Þ=ð2g=aÞ ð6:26Þ
This technique is clearly approximate since although no effects will
return to the boundary from features along the pipeline such as
changes of cross-section, bifurcations, valves, etc. there will be a
more gradual and continuous change due to the action of pipeline
resistance. If necessary, some allowance can be made for this effect by
slowly modifying the value of Jo over time, producing a quasi non-
reflecting boundary. A modified technique which includes the effect
of resistance in the system outwith the modelled area can be found in
Chapter 20.
Non-reflecting boundaries can be used in a variety of contexts. Two
such boundaries can be employed on either side of a booster pumping
station for instance, as depicted in Fig. 6.13. The non-reflecting
boundary can be used in many circumstances. For instance, consider
the case of branch pipe network connected to a larger aqueduct
system. If only events within the branch network are of interest then
non-reflecting boundaries could be installed a short distance along
the aqueduct on each side of the connection point. Alternatively, the
junction itself could be made non-reflecting using the same principle,
by introducing the initial steady flow values of velocity and head in
the aqueduct upstream and downstream of the junction, so that head
at the junction Hj becomes:
Hj ¼ ðJb Ab þ ðVo Ab =Aa þ g=aa Ho ÞAa
ð0Ab =Aa g=aa Ho ÞAa Þ=ð2Aa =aa þ Ab =ab Þ
80
Boundaries
d/s piezometric line
u/s piezometric line
Hd/s
Hu/s
Booster
pumping
station
Non-reflecting boundary Non-reflecting boundary
Dx Dx
C– characteristics from u/s C– characteristics from d/s

pipeline with invariant value J+o pipeline with invariant value J–o
Dt
J+o = Vu/s(o) – g/aHu/s(o) J–o = Vd/s(o) – g/aHd/s(o)

Normal schematisation
between non-reflecting boundaries
Fig. 6.13. Non-reflecting boundaries at a pumping station
or
Hj ¼ ðVo Ab þ 2g=aa Ho Aa Jb Ab Þ=ð2Aa =aa þ Ab =ab Þ
¼ ðconstant Jb Ab Þ=ð2Aa =aa þ Ab =ab Þ ð6:27Þ
where constant ¼ Vo Ab þ 2g=aa Ho Aa .
There are limitations to the application of the above approach.
Clearly the period of analysis has to be limited to < 2L=a where this
is the reflection time from the non-reflecting boundary to the first
source of significant reflection in the pipeline system outwith the area
of interest. The use of this type of boundary is also approximate in
the sense that reflections are continually being transmitted back from
81
the external system through the action of pipeline resistance. This

causes the external value propagating along the characteristic to be
quasi-invariant whereas the above analysis assumes invariant condi-
tions throughout.
6.8 Other bifurcation conditions
6.8.1 Bifurcation with operating valves

From time to time configurations may arise in which there is a choice
regarding how to model the feature. Consider the arrangement shown
in Fig. 6.14 which depicts a bifurcation. Just downstream of the
branch, valves are placed on each of the pipelines 1 and 2. These
valves may operate independently, one may be opening and the
other may be closing for instance. Assuming that there is negligible
separation between the bifurcation point and each of the valves,
then after each time step solution for the six unknowns H1 , H2 , H3 ,
V1 , V2 and V3 should be obtained together. The equations necessary
Piezometric levels
H1
Valve 1
Area = A1
V2
V1 Valve 2 Area = A2
V3
Area = A3
Common horizontal datum H2 H3
Fig. 6.14. Bifurcation with operating valves
82
Boundaries
are:
V1 g=a1 H1 ¼ J1 for pipe No. 1
H1 H2 ¼ KL1 V22 =ð2gÞ for valve No. 1
H1 H3 ¼ KL2 V32 =ð2gÞ for valve No. 2
A1 V1 A2 V2 A3 V3 ¼ 0 at the bifurcation
where KL1 and KL2 are head loss coefficients for valves No. 1 and No. 2
respectively. This set of relationships contains two quadratic equations
and cannot be solved explicitly. An iterative solution is necessary for
each time step. Ellis and Tint (1976) described a method of solving
these equations.
An alternative to solving the complete set of equations together
using a specially constructed routine is to compromise and separate
the valves from the branch so that small increments xi are introduced
between the bifurcation and each valve with a corresponding small time
step t. Beyond each valve and on pipe No. 1, boundaries of the type
described in section 6.6 can be introduced allowing larger distance and
time increments to be used in the rest of the network. In this way
solutions can be found for the six unknowns using existing routines
and without introducing significant inaccuracies.
6.8.2 Isolating valves

Distribution systems will usually be fitted with sluice valves on each
pipeline where it meets a connection point such as is illustrated in
Fig. 6.15. It may be necessary to analyse a number of conditions each
with certain parts of the distribution system isolated, with some of
the sluice valves closed. Rather than have to reconfigure the model
to suit each configuration, a single computation procedure can be
used to analyse all valve arrangements at a connection. Each valve at
a connection can be assigned a code number ki which is set to either
0 or 1. In the situation shown in Fig. 6.15, the valve on line 1 is shut
so k1 ¼ 0 while k2 , k3 and k4 are all 1. The solution for common
junction head H at the connection is:
X X
H ¼ 1=g ðki Ai Ji Þ= ðki Ai =ai Þ ð6:28Þ
83
Piezometric line @ shut valve
H1
Piezometric line @ open valve
H4 H2 H3
A1 Shut valve
A2
A4
A3
Open valve
Common horizontal datum
Fig. 6.15. Isolating valves at a pipe junction
with the summation over all pipes. The head at the valve on each pipe is
then:
Hi ¼ H if ki ¼ 1 or Hi ¼ ai =g Ji if ki ¼ 0
Velocity at the valve on each pipeline i is then:
Vi ¼ Ji ðÞg=ai Hi
6.9 Continuous drawoff

Within a distribution network there may be a large number of discrete
drawoff points — so many that it is not practical to model each point.
Options are to ‘lump’ a group of drawoffs into a single equivalent
large demand point. An alternative is to represent a set of demand
points along a pipeline as a continuous drawoff with distance. With
84
Boundaries
Piezometric line
H Demand points
zi
z Pipeline
Fig. 6.16. Distribution main with demand points
p the discharge qi at any demand point i can be

reference to Fig. 6.16,
written as qi ¼ ki P p hi ¼ H z zi . Along a pipeline total
hi , with
drawoff qt is qpt ¼ P ðki p hi Þ. The equivalent linear drawoff may be
written, q ¼ k h ¼ ðki hi Þ=L, with LP being the length of pipeline.
Taking an average value of hi then, k ¼ ki =L.
Equation (2.4) representing conservation of mass within an element
of pipeline becomes:
@V=@x þ 1=f@=@t þ V @=@xg þ 1=Af@A=@t þ V @A=@xg
p
þ k=A h ¼ 0 ð6:29Þ
The quasi-invariant relationships now become:
ð
p
V g=aH ¼ J ðÞka=A h dt ð6:30Þ
The solutions for V and H at a pipeline node can easily be found.

Routines for solving the quasi-invarient equations with linear drawoff
along each pipeline, at pipe junctions where three or more pipes
meet, can also be set up.
The new term describing the drawoff effect can be approximated
along each segment of characteristic path. If there is no demand on a
pipeline then k ¼ 0 for that line. During a period of low demand k
can be set to a small value with k reaching a maximum during times
85
Start of ring main
Distribution area Distribution area
Area 50R
DN 300 ring main
Pressure vessel
PSV Area 119R–122R
Al Saad water storage compound

and pump house
Area 49R
Distribution area
Blank flange
Air valve at high point on ring main
Fig. 6.17. Typical distribution system
of peak demand. Figure 6.17 shows a pumped distribution system which

was modelled using this linear drawoff approach. Other applications
include irrigation systems where there are a large number of small
demand points.
86
7
Valve closure in a simplified
system
The ‘elastic’ or compressible flow approach developed in the earlier

chapters allows more realistic values of inertial head rise to be estab-
lished. The price to be paid for these more acceptable head changes
is the development of elastic oscillations or vibrations within the pipe-
line system. These oscillations will generally persist until long after the
initial cause of the transient, for example a valve closure has been
completed. This chapter introduces the reader to the concept of
these vibrations in a pipeline system by examining an idealised simple
system.
A circumstance which has been used in a number of publications,
such as Streeter and Wylie (1967), Fox (1977) and Thorley and
Enever (1979), to illustrate the formation of elastic oscillations is the
case of a simple frictionless pipeline of uniform cross-section and
length L which runs between a reservoir and a fast-closing valve at
the opposite end of the line (Fig. 7.1). The positive flow direction is
assumed to be from the upstream reservoir to the valve.
7.1 Instantaneous valve closure at t ¼ 0

Consider the body of liquid within the pipe to be made up of a series
of thin slices as shown in Fig. 7.2. Assuming the valve can be closed
in a short time, almost instantaneously, then the velocity of flow at
the valve face will be reduced to zero over the very short time of
closure. According to the method of characteristics, the invariant
value travelling towards the valve and arriving at the instant of
closure is:
Vo þ g=aHo ¼ Jþ
87
Level = Ho Initial constant piezometric level = Ho

M
Valve
Vo
+X
Length = L
t = 4L/a (seconds)
Time +
J+ = Vo + g/aHo
t = 3L/a (seconds)
J– = Vo – g/aHo
t = 2L/a (seconds)
H = Ho
J+ = –Vo – g/aHo
t = L/a (seconds) V=0
J– = –Vo – g/aHo
t = 0 (seconds)
C+ characteristics
J+ = Vo + g/aHo
Fig. 7.1. Simple pipeline with fast-acting valve
at the valve, V ¼ 0, so that:

Jþ ¼ g=aH (at the valve) or, H (at the valve)
¼ ðVo þ g=aHo Þ=ðg=aÞ ¼ Ho þ Hi
where Hi ¼ aVo =g is the inertial head rise.
88
Valve closure in a simplified system
Propagation rate of pressure wavefront = a
Inertial head rise = aVo/g

Steady flow piezometric line
Static ‘layers’
Steady flow velocity
Vo
‘Layers’ travelling towards valve
Valve abruptly closed
Additional travel of
first and second layers
Fig. 7.2. Representation of pressure wave progress
At the instant of closure only a very thin layer of liquid, immediately

against the upstream face of the shut valve, will be brought to rest
ðV ¼ 0Þ and will be under the increased head ðHo þ Hi Þ.
7.2 From 0 < t L=a

A short time later a further thin layer of liquid will be decelerated as it
meets the now stationary layer of liquid against the valve. Both layers
are not decelerated simultaneously as the second layer now has a
short distance to travel since the first layer is occupying a shorter
length of pipe (Fig. 7.2) on account of:
(a) having been compressed into a smaller volume under the increased
head Hi
(b) since the pipe wall has itself been distended giving an increased
cross-sectional area to accommodate the layer of stationary
liquid.
When the second layer of liquid is brought to rest its velocity is now
zero and pressure within the layer has increased by the inertial head rise
89
Hi . Subsequent layers of liquid will be brought to rest in a similar

manner so that at some time t after valve closure the head rise Hi
will have reached a point x upstream of the valve. Since generally
V a then dx=dt a and x=t ¼ a. The invariant J propagating
upstream in the ve direction will have a value provided by the flow
conditions at the valve just after closure, that is:
J ¼ V g=aH or J ¼ 0 g=aðHo þ Hi Þ ¼ g=aHo Vo
The wave travelling upstream in the pipeline produces an increase in
internal pressure and is known as a ‘compression wave’. After a time
L=a the ‘pressure wave’ front will have reached the reservoir end of
the system. Characteristic paths as the wave travels upstream and is
then reflected back towards the valve are also shown in Fig. 7.1.
Up until time L=a, water has continued to flow into the pipeline from
the reservoir at velocity Vo . At this time the entire liquid column within
the pipeline is now at rest ðV ¼ 0Þ under the head Ho þ Hi . The extra
volume of liquid now contained in the pipeline under the increased
pressure head Hi is:
Volume ¼ Vo AL=a
The state where the pressure head just inside the pipeline is greater
than the reservoir head by the amount Hi is evidently unstable.
Again considering the characteristic arriving at the reservoir at this
time, then:
J ¼ g=aHo Vo
At the reservoir H ¼ Ho (constant), therefore:
J ¼ V g=aHo ¼ g=aHo Vo ¼ g=aðHo þ Hi Þ
or
V ¼ g=aHi ¼ Vo
7.3 L=a < t 2L=a

After the pressure wave reaches the upstream end of the pipeline, liquid
starts to flow back into the reservoir from the pipeline with velocity Vo
and head is reduced back to its original value Ho . This effect initially
influences only the layer of liquid adjacent to the reservoir, but sub-
sequently successive layers of liquid are affected with a pressure wave
travelling along the pipeline from the reservoir towards the valve
with characteristic velocity, dx=dt ¼ a and producing a reduction in
90
internal pipeline pressure. A wave which causes a reduction in pressure

is known as a ‘rarefaction’ wave.
The invariant travelling in the positive direction of flow along the
Cþ characteristic from the reservoir to the shut valve has a value
given by the conditions just found at the reservoir. Thus:
Jþ ¼ V þ g=aH ¼ Vo þ g=aHo
After a total time 2L=a has elapsed from the initial valve closure, this
reflected pressure wave has now reached the shut valve. The invariant
Compression wave
Ho a Hi
M
V = Vo
V=0
(a) 0 £ t £ L/a
Rarefaction wave
Ho a Hi
M
V = –Vo V=0
(b) L/a £ t £ 2L/a
Rarefaction wave
Ho
M
a V=0 –Hi
V = –Vo
(c) 2L/a £ t £ 3L/a
Ho Compression wave
M
a V=0 –Hi
V = +Vo
(d) 3L/a £ t £ 4L/a
Fig. 7.3. Pressure wave front at different stages following valve closure
91
value arriving at the valve face at this time remains constant at:
Jþ ¼ Vo þ g=aHo
At the valve face V ¼ 0 so that:
Jþ ¼ Vo þ g=aHo ¼ 0 þ g=aH
or
H ¼ Ho aVo =g ¼ Ho Hi
From initial valve closure at time ¼ 0 until time ¼ 2L=a, when a
reflected pressure wave returns to the valve from the reservoir, piezo-
metric level at the valve remains at Ho þ Hi . After time ¼ 2L=a,
piezometric level at the valve has fallen to Ho Hi . This time 2L=a
from the initiation of the pressure transient is the minimum time
before any ‘relief of pressure’ occurs at the source of the hydraulic
transient. It is sometimes referred to as the ‘critical period’. Remaining
with this simple example, it follows that any closure time which is
2L=a will produce the same maximum head rise as an instantaneous
valve closure. Closure times >2L=a will start to bring benefits of pres-
sure relief with consequent maximum inertial head rise Hi being
<aVo =g. Figure 7.3 shows the position of the wavefront at different
stages as it travels to and fro in the pipeline.
7.4 2L=a < t 3L=a

After this reflected pressure wave has reached the shut valve, head
against the valve has to fall by an amount aVo =g below the static
level Ho in order to ensure that flow remains at rest against the valve
face. As before, this change initially affects only the layer of liquid in
contact with the valve but, as time passes, successive layers of liquid
are influenced as the rarefaction wave travels upstream. Consider the
J Riemann invariant propagating along the C characteristic. Its
value is given by conditions at the valve. Thus:
J ¼ V g=aH ¼ 0 g=aðHo Hi Þ ¼ Vo g=aHo
The characteristic arrives at the reservoir a total time 3L=a after initial
valve closure. At this stage the entire pipeline liquid column is at rest at
a piezometric level Ho Hi . With the reservoir at a level Ho , conditions
across the inlet to the pipeline are again unstable.
Consider the Riemann invariant value at the reservoir after time
¼ 3L=a. At the reservoir, H ¼ Ho , therefore:
J ¼ V g=aHo ¼ Vo g=aHo or V ¼ Vo
92
7.5 3L=a < t 4L=a

After time 3L=a the velocity at the reservoir end of the pipeline
becomes þVo . This is the Jþ invariant value that existed initially
before the valve was shut. So the layer of liquid adjacent to the reservoir
is accelerated back to its original value of Vo and the piezometric level is
restored to its initial value of Ho . A short time later the next layer of
liquid is also accelerated to Vo and its piezometric level increases to
Ho as a compression wave starts to travel downstream towards the
closed valve.
The Cþ characteristic travelling from the reservoir to the shut valve
has a Jþ invariant value:
Jþ ¼ Vo þ g=aHo
As time passes the pressure wave propagates downstream to reach the
valve after time 4L=a from the initial valve closure. When the pressure
wave reaches the shut valve the entire pipeline is under the reservoir
head Ho and velocity is everywhere Vo directed towards the valve.
Conditions are thus precisely the same as existed in the pipeline at
the instant the valve started to close. In the absence of any damping
mechanism, propagation of pressure waves within the pipeline will
continue to be repeated at successive time intervals 4L=a till infinity
as illustrated in Fig. 7.4. The time interval 4L=a is the period of
elastic oscillations within this pipeline.
In practice there is damping in the shape of pipeline resistance and
‘minor’ losses, as well as entry loss and kinetic energy development at
U/s of valve
+aVo/g Close to reservoir
–aVo/g
Mid-point
Ho of pipeline
Datum 0 L/a 2L/a 3L/a 4L/a 5L/a 6L/a 7L/a
Time
Critical time = 2L/a
Period of oscillation = 4L/a
Fig. 7.4. Time histories of inertial head in pipeline after valve closure
93
the inlet to the pipeline from the reservoir. These effects will pro-
gressively erode the transient event until eventually a new steady
flow regime prevails. In theory there will always be some small residual
vestige of the transient with truly steady flow never actually being
reached. In practice the flow may be declared steady when amplitude
of the residual oscillation has become smaller than some selected
finite value.
It will have been noted that compression waves have travelled in
both the positive and negative directions of motion within the pipeline.
The same applies to rarefaction waves. Direction of wave motion, for
example in the positive x direction, does not imply that the wave
produces an increase in pressure.
94
8
Actual pipelines
Whereas Chapter 7 examined an idealised valve closure condition, the

present chapter illustrates some of the differences in behaviour which
can occur between a simplified pipeline and actual systems. Obvious
differences which spring to mind are the effects of pipeline resistance
and losses generally which are a necessary factor if an eventual
almost steady flow condition is to be established. Other aspects
include the possible development of sub-atmospheric pressures and
the radical effect which appearance of gas or vapour cavities can
produce.
8.1 Attenuation
Before proceeding to consider specific examples, some discussion is
appropriate regarding the phenomenon of attenuation which occurs
in pipelines generally but whose existence may not be obviously
apparent except where the pipeline is relatively long. Consider the
circumstance of a valve closed abruptly at the downstream end of a
long pipeline having uniform properties as shown in Fig. 8.1. Steady
flow velocity is Vo , pipe diameter is D, uniform wave speed is a and
dx=dt ¼ a for simplicity. Friction factor f is taken to be constant for
convenience and the effects of pressure change on velocity under
steady flow are ignored. Uniform gradient of piezometric line is So
under the original steady flow. Datum for head measurement is taken
to be at the level of the steady flow piezometric line at the valve as
shown in Fig. 8.1.
Head rise Hi just upstream of the valve when the valve is just closed at
time to ¼ 0 is aVo =g. At some time t after closure, the compression wave
front caused by valve closure will have propagated upstream to a point
95
V = Vo
Transient piezometric line or
hydraulic gradient
Hv
Hx
Compression wave
H
Steady flow
Hi = aVo/g
Piezometric line
Gradient = So
Horizontal datum
Initial steady
velocity = Vo
X
L
C– characteristic
J– = –Vo – g Ú S dt
J+ = Vo + gSot
t
C+ characteristic
J+ = Vo + gSot/2
t/2
x/2
Time to
x
Fig. 8.1. Attenuating wave front
96
Actual pipelines
x ¼ at. If velocity is reduced to zero as the wave passes a point, with

inertial head rise remaining at aVo =g then total head downstream of
the wave front will rise continuously at a rate So a. Any increase in
head downstream of the wavefront must be accompanied by a continued
flow downstream of the wavefront and so velocity across the front cannot
be entirely reduced to zero and inertial head rise must be <aVo =g. A
simple analysis is presented to demonstrate this effect.
With reference to Fig. 8.1, the backward C characteristic starting
from the valve at to will have a quasi-invariant J with a value at
time t of:
ð
J ¼ Vo g S dt ð8:1Þ
where S is the resistance gradient.
8.1.1 Conditions at the wavefront

Just upstream of the wavefront at any time t the Cþ characteristic will
carry the quasi-invariant Jþ value:
Jþ ¼ Vo þ g=aSo at ¼ Vo þ gSo t ð8:2Þ
At intersection of these opposing paths at time t then:
ð
1
V ¼ ð Jþ þ JÞ=2 ¼ Vo þ gSo t Vo g S dt
2
ð
¼ g=2 So t S dt ð8:3Þ
and
ð
1
H ¼ ð Jþ JÞ=ð2g=aÞ ¼ a=g Vo þ gSo t Vo g S dt
2
ð
¼ aVo =g þ a=2 So t þ S dt ð8:4Þ
Total head H is made up of the initial steady flow head plus the inertia
head rise Hx , so that Hx ¼ H So at or Hx ¼ H So x. Then:
ð
Hx ¼ aVo =g þ a=2 So t þ S dt So at
ð
¼ aVo =g þ a=2 S dt So t ð8:5Þ
97
Using the Darcy equation to evaluate head loss due to friction then,
So ¼ f Vo2 =ð2gDÞ and S ¼ f V 2 =ð2gDÞ. Substituting in the expression
for velocity V at the wavefront:
ð
2 2
V ¼ f=ð4DÞ Vo t V dt ð8:6Þ
If it is assumed that velocity varies linearly with time along the C

characteristic, then dV=dt ¼ V=t ¼ constant or dt ¼ dV=constant so
that:
ð ð
V 2 dt ¼ 1=constant V 2 dV
giving:
V ¼ ft=ð4DÞðVo2 V2 =3Þ
or, solving for V:
p
V ¼ 6D=ð ftÞð f1 þ 3½Vo ft=ð6DÞ2 g 1Þ ð8:7Þ
A simpler expression for V can be found by assuming that
3½Vo ft=ð6DÞ2 1. Using a series expansion and truncating at the
first term:
V Vo2 ft=ð2DÞ ¼ Vo2 fx=ð2aDÞ ¼ g=aSo x ð8:8Þ

Thus velocity just downstream of the wavefront increases with distance
upstream.
Turning to the equation for Hx :
ð
Hx ¼ aVo =g þ a=2 S dt So t
and making similar substitutions for S and So then:
Hx ¼ a=g½Vo þ ft=ð4DÞðV 2 =3 Vo2 Þ

but V ¼ ft=ð4DÞðVo V 2 =3Þ so that as expected:
Hx ¼ a=gðVo VÞ ð8:9Þ
This equation demonstrates that the inertial head rise at the front of the
pressure wave diminishes linearly with distance upstream for the
assumed linear variation in V.
98
Actual pipelines
8.1.2 Conditions when the wave height is of zero amplitude

Returning to the expression for velocity just downstream of the
wavefront,
p
V ¼ 6D=ðftÞf ð1 þ 3½Vo ft=ð6DÞ2 Þ 1g ð8:7Þ
When the wavefront has been attenuated to the point when V ¼ Vo
then,
p
Vo ft=ð6DÞ þ 1 ¼ ½1 þ 3ðVo ft=ð6DÞÞ2
Setting y ¼ Vo ft=ð6DÞ and squaring,
y2 þ 2y þ 1 ¼ 1 þ 3y2
rearranging then,
2y2 2y ¼ 0 or y ¼ 1
giving,
Vo ft=ð6DÞ ¼ 1 or t ¼ 6D=ðVo fÞ
Since x ¼ at therefore,
x ¼ 6aD=ðVo fÞ ð8:10Þ
The inertial head rise Hx at the wavefront is given by the expression,
Hx ¼ a=g½Vo þ ft=ð4DÞðV 2 =3 Vo2 Þ
When V ¼ Vo and t ¼ 6D=ðVo fÞ then,
Hx ¼ a=g½Vo þ ft=ð4DÞðVo2 =3 Vo2 Þ
¼ a=gðVo ft=ð4DÞ 23 Vo2 Þ
or as expected,
Hx ¼ a=g½Vo f6D=ð4DfVo Þ 23 Vo2 ¼ 0 ð8:11Þ
Confirming that the wave front is of zero height when V ¼ Vo .
To gauge the magnitude of x consider the following values: Vo ¼ 1 m/s;
f ¼ 0:02; D ¼ 1 m and (i) a ¼ 1200 m/s and (ii) a ¼ 300 m/s. Sub-
stituting in equation (8.10), then (i) x ¼ 120 km and (ii) x ¼ 30 km.
Only in very long pipelines is it likely that the wave will be completely
attenuated. In shorter lines significant attenuation can still be
observed.
99
8.1.3 Conditions at the closed valve

Total head just downstream of the wavefront at time t is given by:
ð
1
H ¼ a=gVo þ S x þ S dx
2 o
¼ a=gVo þ fx=ð4gDÞðVo2 þ V 2 =3Þ
At the shut valve at time t the Jþ quasi-invariant has the value,
ð
Jþ ¼ Vo þ g=aSo x=2 g S dt
Ð
where the integral g S dt is evaluated between t=2 and t. Making
substitutions for So and S then in terms of velocity V at the wavefront,
Jþ ¼ Vo þ ft=ð4DÞðVo2 V 2 =12Þ ð8:12Þ
and head at the closed valve is thus,
Hv ¼ a=gVo þ fx=ð4gDÞðVo2 V 2 =12Þ ð8:13Þ
The corresponding head at the wavefront at time t is,
H ¼ a=gVo þ fx=ð4gDÞðVo2 þ V 2 =3Þ
Comparing the expressions for Hv and H it can be seen that the head at
the valve lags behind head at the wavefront by an amount,
H Hv ¼ 5fxV 2 =ð48gDÞ ð8:14Þ
8.1.4 Conditions downstream of a pump or valve

This phenomenon of attenuation is not confined to the pipeline
upstream of a terminal discharge valve: attenuation of a rarefaction
wave also occurs downstream of an in-line valve. In general, such
behaviour is to be anticipated in pipelines connected to a pumping
station following a pumping failure.
8.2 A uniform gravity main

Consider the relatively straightforward gravity main shown in Fig. 8.2a.
The main is of ductile iron (DI) with a length of 19.757 km and nominal
diameter (DN) 450. Pipeline gradient is uniform throughout its length
at 1 :398. Water is pumped into the upstream reservoir from a set of
boreholes. From this reservoir, water then flows to a second lower
100
Actual pipelines
Compression wave
160
140 t = 18.41 s
120
t = 13.81 s
100 t = 9.2 s
t = 4.6 s
Head (mAD)
80
60
Steady flow profile
t=0s
40
Downstream valve
Pipeline profile
20 Um Ghafa Water Project
3 s closure of downstream valve
instantaneous hydraulic profiles 4.6 £ t £ 23.01 s
0
0
790.3
1580.6
2370.8
3161.1
3951.4
4741.7
5532
6322.2
7112.5
7902.8
8693.1
9483.4
10 273.6
11 083.9
11 854.2
12 644.5
13 434.8
14 225
15 015.3
15 805.6
16 595.9
17 386.2
18 176.4
18 986.7
19 757
Chainage (metres from upstream reservoir)
Fig. 8.2a. Propagation of compression wave
180
160 Rarefaction wave
140
120 t = 27.61 s t = 32.21 s

Head (mAD)
100 Pipeline profile Rarefaction wave t = 36.82 s
80
t = 46.02 s t = 41.42 s
60
40 Closed valve
Um Ghafa Water Project
20 3 s closure of downstream valve
instantaneous hydraulic profiles
27.61 £ t £ 46.02 s
0
0
790.3
1580.6
2370.8
3161.1
3951.4
4741.7
5532
6322.2
7112.5
7902.8
8693.1
9483.4
10 273.6
11 083.9
11 854.2
12 644.5
13 434.8
14 225
15 015.3
15 805.6
16 595.9
17 386.2
18 176.4
18 986.7
19 757
Fig. 8.2b. Rarefaction waves following reflection
101
80
70
t = 78.23 s
Pipeline profile
60
50
Head (mAD)
t = 73.63 s
t = 55.22 s
40
t = 59.83 s
30
t = 69.03 s
20
t = 59.83 s
Um Ghafa Water Project Closed valve
10 3 s closure of downstream valve t = 55.22 s
55.22 £ t £ 78.23 s
0
0
790.3
1580.6
2370.8
3161.1
3951.4
4741.7
5532
6322.2
7112.5
7902.8
8693.1
9483.4
10 273.6
11 083.9
11 854.2
12 644.5
13 434.8
14 225
15 015.3
15 805.6
16 595.9
17 386.2
18 176.4
18 986.7
19 757
Fig. 8.2c. Progressive collapse of vapour cavities
120
t = 96.64 s
100 t = 115.05 s
t = 87.44 s
80
Head (mAD)
t = 119.65 s
60
t = 82.84 s
40 Pipeline profile
20 Um Ghafa Water Project

3 s closure of downstream valve Closed valve
82.84 £ t £ 119.65 s
0
0
790.3
1580.6
2370.8
3161.1
3951.4
4741.7
5532
6322.2
7112.5
7902.8
8693.1
9483.4
10 273.6
11 083.9
11 854.2
12 644.5
13 434.8
14 225
15 015.3
15 805.6
16 595.9
17 386.2
18 176.4
18 986.7
19 757
Fig. 8.2d. Rising head after cavity removal
102
Actual pipelines
downstream reservoir within a military compound with the rate of

flow in the main being controlled using a gate valve just upstream of
the military reservoir.
The wellfield contained an initial set of 30 boreholes, it being the
intention to expand the size and capacity of this field in the future.
Diameter of the 20 km gravity main was chosen to satisfy the
maximum final output from an expanded wellfield. During the first
phase of development it was necessary to operate with the downstream
valve partially closed in order to limit flow in the pipeline to the
first-stage wellfield output of 125 litres/s. The hydraulic gradient, or
piezometric line, for this initial stage was as shown in Fig. 8.2a.
When a valve is being closed from the fully opened position, for a
substantial part of the valve stroke there is little effect upon flow and
head conditions and it is only over the final 15—20% of movement
that substantial deceleration occurs. In the present example however,
the valve was initially set at only 4.5% open and when the valve
closure occurs from this position it will impose important flow and
head variations on the water from the onset of movement.
To illustrate hydraulic transient behaviour, consider events during
and after a valve closure lasting just 3 s. The corresponding time to
close the valve from fully opened would be 67 s.
The compression wave generated by closure travels upstream
through the water-filled pipeline at an acoustic velocity calculated to
be 1113 m/s, with the front of the pressure wave reaching the upstream
reservoir after 17.75 s — that is, L=a seconds from commencement of
valve closure. Figure 8.2a shows successive positions of this wave
front between times 4.6 s and 18.41 s, by which time the wave has
reached the reservoir and commenced to be reflected.
It will be noted that head rise at the valve does not cease after the
final valve closure at 3 s but continues to increase, albeit more slowly.
This is due to the initial pressure wave rise being ‘built’ onto the
initial steady flow hydraulic gradient so that piezometric level just
upstream of the valve increases due to attenuation as described in
the previous section.
After reflection from the upstream reservoir, the reflected rarefaction
pressure wave front travels to the now closed valve which it reaches at
2L=a seconds when a relief of pressure starts to occur. Successive
positions of this reflected wave are shown in Fig. 8.4b at times of
27.61 and 32.21 s. On arriving at the valve, the rarefaction wave
causes piezometric level to fall at time 36.82 s until sub-atmospheric
pressure and eventually vapour pressure occurs. Head at the valve
103
now ceases to fall. After 2L=a seconds the reflected wave from the valve
travels upstream, reducing pressure along the majority of the pipeline as
shown in Fig. 8.2b at times 41.42 and 46.02 s. However, head does not
remain at vapour pressure near to the valve. Instead an initial hydraulic
gradient equal to the slope of the pipeline is established along these
parts of the line subject to vapour pressure. A quasi free surface flow
is established towards the valve, causing pressure to increase gradually
at the valve and in adjoining parts of the pipeline as shown in Fig. 8.2c
at times 55.22, 59.83 and 69.03 s.
When the rarefaction wave reaches the upstream reservoir after time
3L=a seconds it is reflected into the pipeline in the form of a compres-
sion wave producing a recovery of head. A situation is developed in
which head is increasing along upstream parts of the pipeline as
shown for times 55.22, 59.83 and 69.03 s. At the same time head is
rising over downstream parts near to the closed valve (Fig. 8.2c),
with vapour bubbles being removed as pressure rises. Eventually these
compression waves meet at a point within the system as the final
vapour cavity is removed and a head rise occurs at this point as
shown in Fig. 8.2c at time 73.63 s.
This head rise then spreads both upstream and downstream, pro-
ducing a more general head rise along the pipeline as illustrated in
180
3 s closure of downstream valve
160 time histories of head
140
120 Upstream of valve

Head (mAD)
100
80
60
8 km from upstream
40 reservoir
20
0
0.354
4.248
8.142
12.036
15.93
19/824
23.718
27.612
31.508
35.4
39.294
43.188
47.082
50.976
54.87
58.764
62.658
66.552
70.446
74.34
78.234
82.128
86.022
89.916
93.81
97.704
101.596
105.492
109.386
113.279
117.173
Time (s)
Fig. 8.3. Head plotted against time for 3 s valve closure
104
Actual pipelines
140
2 min closure of downstream valve
120 time histories of head
100
Head (mAD)
80
60
8 km from upstream
40 reservoir Upstream of valve
20
0
0.354
7.06
13.806
20.532
27.258
33.964
40.71
47.436
54.162
60.888
67.614
74.34
81.066
87.792
94.518
101.244
107.97
114.695
121.421
128.147
134.873
141.599
148.326
155.052
161.778
166.504
175.23
181.956
188.682
195.400
202.134
208.80
Time (s)
Fig. 8.4. Head plotted against time for 2 min valve closure
Fig. 8.2d at times 82.84, 87.44 and 96.64 s. Thereafter head starts to
decline as shown for times 115.05 and 119.65 s.
In deciding on an appropriate time of valve closure it is important to
consider not only the initial head rise at the valve on closure but also
the potential for unacceptable pressures being developed following
reflection of the initial pressure wave.
Figure 8.3 shows predicted variations of piezometric level at chainage
8 km from the upstream reservoir and just upstream of the valve for the
3 s valve closure. Figure 8.4 shows corresponding predictions for a more
extended valve closure lasting 2 min.
105
9
Valve operations
The term valve encompasses many different types, each with a wide
range of duties. Valves discussed in this chapter are those located
on a pipeline and through which flow passes during normal opera-
tion. These are often used to control discharge and/or pressure.
Many valves are under operator control using actuators to open and
close the valve at a pre-defined rate. Others may be manually oper-
ated using a handwheel. Still other valves are fitted with control
systems which respond to changes in flow and pressure throughout
the day. These valves may be set to maintain a set upstream or
downstream pressure or to produce a set range of downstream
pressure and flow. Each of these valve patterns has the capability to
initiate hydraulic transients during the process of flow adjustment.
Where a valve is not moving it will merely act as a partial trans-
mitter/reflector of surge effects arising elsewhere in the pipeline
system.
This chapter includes several examples of gravity pipeline systems
in which flow is controlled using valves. Valve operation is one of the
primary sources of pressure transient behaviour. In the majority of
cases discharge is controlled by a valve at the downstream end of the
main. A valve in this position may be called a ‘terminal discharge
valve’. Pressure surges will be developed in the line upstream of the
valve as it is opened or closed. In some instances a valve may be
positioned at some point within a pipeline system with appreciable
lengths of line both upstream and downstream of the valve. A valve
in this position may be called an ‘in-line valve’ with transient effects
produced both upstream and downstream of the valve. Other valves
are the discharge valves just downstream of pumps and the isolating
valves distributed along long pipelines.
106
Valve operations
9.1 Treated water main

The first example concerns a steel gravity main, diameter 1000 mm and
length 16.8 km, with a radically different profile to the uniform profile of
the example in Chapter 8. The pipeline profile undulates considerably
(Fig. 9.1). The water in this pipeline has been treated and so it is a
requirement that minimum pressure should remain positive
throughout. Inflow to the lower reservoir is normally controlled using
float-operated valves which respond to changing level in the receiving
reservoir. In addition, emergency butterfly valves are provided upstream
of the float valves. The function of the butterfly valve is to close in the
event that the float valve does not shut.
Maximum transient pressure caused by butterfly valve closure should
not exceed the peak pressure developed during float valve closure.
Maximum piezometric level

during float valve closure
140
130
TWL Flow rate = 95.47 Mld when
upstream reservoir is at
120 top water level (TWL)
LDO
110
Flow rate = 90.92 Mld when
100 upstream reservoir is at
lowest draw-off (LDO)
90
Head (mAD)
80
70
60
50
40
30
Pipeline profile
20
10
0
0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17
Chainage (km)
Fig. 9.1. Steel gravity main
107
10 000.0
Valve head loss characteristics
1000.0
100.0
Valve head loss coefficient, KL
Submerged discharge
10.0 Butterfly
Gate
Needle
1.0
0.1
0.01
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0
Valve position as a proportion of port stroke
Fig. 9.2. Valve loss coefficients plotted against stroke
The head loss characteristic of the butterfly valve and of some other
valves was as shown in Fig. 9.2, where head loss H ¼ KL V 2 =ð2gÞ. It
will be noted that changes in KL are modest for much of the stroke
from open to shut. Only when the valve has become about 15%
closed does the value of KL start to change more rapidly. Unfortunately
the availability of head loss data is more limited towards the more
interesting closed valve end of the stroke. Typically, data curves
obtained from manufacturers terminate around the 10% open position
and it is often necessary to complete the curve down to the closed
108
Valve operations
125
Start of valve operation
120
115
Head (mAOD)
Mid-point of main
110
Upstream of
butterfly valve
105
Head variations during 1 min

100 opening of butterfly valve
95
0.365
3.285
6.205
9.125
12.045
14.965
17.885
20.805
23.725
26.645
29.565
32.485
35.405
38.325
41.245
44.165
47.085
50.005
52.925
55.845
58.785
61.685
64.605
67.525
70.445
73.365
76.285
79.205
82.125
85.045
Time (s)
Fig. 9.3. Head variations during valve opening
position. As the valve door approaches its seat KL ! 1 and in compu-

tations a shut valve coefficient of the order of 1012 is required to reduce
velocity effectively to zero.
Maximum flow rate through the gravity main is dependent upon the
upstream reservoir level while discharge head takes place above the
water level in the downstream service reservoir. Flow rate varied from
90.92 Mld to 95.47 Mld over the range in upstream reservoir elevation.
While primary interest usually centres upon valve closure and the
transient pressures developed as a consequence, it is also advisable to
check pressures occurring during the process of valve opening.
Performing a valve opening analysis avoids the need to establish an
initial steady flow rate and hydraulic gradient. Instead simple initial
static conditions can be set and the valve opened to permit flow to
become established. Figure 9.3 shows the predicted head variation
just upstream of the valve and at the halfway point along the main
for a one-minute valve opening interval. It can be seen how rapidly
the head can fall as the valve is opened over the first 10% or so of its
stroke. In contrast, flow in the pipeline at the valve is predicted to
develop in a stepped manner over an interval of about 4 min (Fig. 9.4).
The rapid fall in head, as the valve starts to open, travels upstream as
a negative pressure wave and it is possible that this wave may cause
109
1.4
At butterfly valve
1.2
1.0
0.8
Velocity (m/s)
0.6
0.4 Development of velocity in DN 1000 main

for 1 min butterfly valve opening
0.2
At upstream reservoir
0
–0.2
0.365
9.125
17.885
26.645
35.405
44.165
52.925
61.685
70.445
79.205
87.965
96.725
105.485
114.245
123.005
131.765
140.525
149.285
158.045
166.805
175.585
184.326
193.086
201.846
210.606
219.366
228.126
236.886
245.647
254.407
263.167
271.926
Time (s)
Fig. 9.4. Flow development after valve opening
operation of air valves at high points along the pipeline. Figure 9.5
shows the curves of minimum and maximum head during this valve-
opening sequence.
With computer simulation carried out for a sufficient length of time,
essentially steady flow conditions will have been established at the end
140
Maximum head
120
100
Minimum head
Head (mAOD)
80
60
40 Pipeline profile
20
Maximum and minimum head in
DN 1000 main 1 min butterfly valve opening
0
0
835
1671
2508
3342
4177
5012
5848
6401
6968
7818
8669
9235
10 095
10 668
11 529
12 390
13 251
14 112
14 701
15 585
16 469
Chainage (m)
Fig. 9.5. ‘Envelope’ curves of head during valve opening
110
Valve operations
200
180
160
140 Head at midpoint of main

Head (mAOD)
120
100
80 Head upstream of valve

Full vacuum upstream of valve
60
40
DN 100 steel gravity main
20 3 min butterfly valve closure
0
0.365
12.045
23.725
35.405
47.085
58.765
70.445
82.125
93.805
105.485
117.165
128.845
140.525
152.205
163.885
175.585
187.246
196.926
210.606
222.286
233.966
245.647
257.327
269.006
280.686
292.368
304.045
315.725
327.405
339.064
350.764
362.444
Time (s)
Fig. 9.6. Head variations for 3 min valve closure
of computations. Within these computations the hydraulic transients

were predicted to decay asymptotically under the action of pipeline
resistance but with a tendency for small oscillations in head and flow
to persist for an appreciable time. It is up to the user to decide when
these small fluctuations are no longer significant.
Closure of the valve from the fully opened position involves move-
ment of the valve door over a substantial part of the stroke when the
value of KL changes only slowly. The corresponding flow and head
changes in the pipeline are modest. Only during the final 10% or so
of closure does KL increase rapidly, with large changes of flow and
head taking place. Figure 9.6 shows changing head upstream of the
valve and at the mid-point of the main for a 3 min closure. Closure
in a time < 2L=a will produce a head rise which is essentially that
produced by an instantaneous closure, ignoring the effect of attenua-
tion. Velocity decreases only slowly over initial parts of the closure
and then increasingly steeply until the valve is shut. Subsequent reflec-
tion of this initial upsurge can produce minimum head conditions which
cause air valves to open. Figure 9.7 shows maximum and minimum
head falling to the pipeline level at several locations along the
undulating profile. Since head rise is a function of the flow deceleration
dV=dt, it can be seen that the peak inertial head rise will depend upon
the maximum deceleration rate, even although this deceleration only
occurs for a relatively small fraction of the closure time.
111
200
180
Maximum head
160 Maximum
Maximum and minimum head during pressure head
140 3 min butterfly valve closure
Head (mAOD)
120
Minimum head
100
80
60
40
20 DN 1000 pipeline profile

0
0
835
1671
2506
3342
4177
5012
5848
6401
6968
7818
8669
9235
10 095
10 668
11 529
12 390
13 251
14 112
14 701
15 585
16 469
Chainage (m)
Fig. 9.7. Envelope curves of head for 3 min valve closure
The simplest way to reduce rates of deceleration and thus inertial

head is through extending the time of closure of the valve. For the
present example, Fig. 9.8 shows maximum pressure upstream of the
valve as a function of closure time. To reduce maximum pressure to
18
+
17
+
16
Pressure (bar g)
15
+
Maximum pressure on
14 float valve closure
+
13
+
12
0 2 4 6 8 10 12 14 16 18
Time (min)
Fig. 9.8. Maximum pressure plotted against valve closure time
112
Valve operations
similar levels obtained during float valve closure requires a valve stroke
time >13 min. Closure intervals can become quite prolonged. This is a
consequence of the need to slow valve movement during the final 10—
20% or so of closure in order to reduce the deceleration rate. This same
slow valve movement also takes place over the larger part of the stroke
when little flow change occurs.
9.2 Improving valve operation

An ideal valve would seek to produce a linear variation of velocity with
stroke. In practice this is very difficult to achieve but some steps can be
taken to improve this relationship by reducing the degree of non-
linearity between velocity and stroke. For example, a reduction in
valve diameter can improve this relationship but at a price. The head
loss through the smaller valve is greater than for a line sized valve so
that maximum flow attainable is reduced.
The flow through a large-diameter pipeline can be split into two or
more smaller-diameter branches before entering a downstream reser-
voir. The valves on each branch will also be smaller. These can be
set to close in sequence so that a more gradual deceleration can be
achieved. It is also easier for an operator to close a smaller valve.
One method of reducing the overall time of valve closure is to utilise
an actuator capable of closing the valve at alternative rates over
different parts of the stroke. For example, it is possible to cause the
valve to close relatively quickly during initial stages of movement, say
from 100% open to 20% open, using 50% of the overall closure time
and to move the valve more slowly from 20% open to shut over the
remaining 50% of closure time. Potentially, using a multi-stage
closure, it would be possible to attain an almost linear deceleration
over the entire valve stroke. An example using two-stage closure is
included in section 9.3.
Common ‘off-the-shelf ’ valves such as gate valves, sluice valves and
butterfly valves have quite unfavourable characteristics as regards surge
control because of the highly non-linear relationship between KL and
the valve position (Fig. 9.2).
9.3 Two-stage valve closure

Release of water from a high dam may use a valve which discharges
to a downstream watercourse or into a supply pipeline. Since a
considerable amount of energy is involved it is important to include
113
Spillway
Penstocks
Valve house
Draw
off
Downstream
Dam dam face
crest
120 m
Draw-off well
Flow
regulating valve
Emergency Valve
Section through dam
butterfly house
valve
Burst site
Fig. 9.9. Reservoir outlet arrangement
an emergency closing valve upstream of the terminal discharge valve.

In the event of a pipe burst or failure of the terminal valve, flow will
accelerate within the outlet pipework with velocities of the order of
30 m/s being attained. Figure 9.9 illustrates outlet works from a high
dam with emergency closing butterfly valves included. These butterfly
valves (Fig. 9.10) are fitted with a lever and weight to assist closure
114
Valve operations
Weight
Lever
Valve door
Flow
Piston and cylinder
Fig. 9.10. Emergency closing butterfly valve
but with an oil-filled piston and cylinder arrangement attached to the

lever limiting rates of movement. The rate of valve closure can be
regulated by altering the orifice through which oil passes while the
valve moves. In this case the orifice arrangement can allow two rates
of movement of the valve. A relatively rapid movement of the valve
was permitted from the 908 position (100% open) down to 98 over
the first 50% of the total closure time. Then a much slower closure
rate was used from 98 open to 08 (shut) for the second 50% of stroke
time.
Closure times from 15 s to 60 s were studied during which the valve
angle changed from 908 (fully open) to 98 in half the time and from 98 to
08 (shut) in the remaining time.
Figure 9.11 shows the variation of velocity during a 30 s closure event
and Fig. 9.12 depicts the changing head upstream of the valve during
closure. It will be noted that maximum head exceeds the static level
by <20 mWG.
Calculations can also be used to predict changing forces and
torques acting upon the valve door. In addition to head loss co-
efficient data, manufacturers can also supply force and torque data as
a function of valve position. This information can be used to establish
the force and its direction, acting on the valve door and from this the
frictional resistance moment offered by the valve shaft as it rotates.
Hydrodynamic torque can also be calculated and when added to the
frictional resistance torque the net closure moment can be found as a
115
30
30 s emergency valve closure
25
20 At base of At emergency
draw-off well butterfly valve
Velocity (m/s)
15
10 Valve 10% open
5 Valve shut
–0.5
0.02
1.04
2.06
3.08
4.10
5.12
6.14
7.16
8.18
9.20
10.22
11.24
12.26
13.28
14.30
15.32
16.34
17.36
18.38
19.40
20.42
21.44
22.46
23.48
24.50
25.52
26.54
27.56
28.58
29.60
30.621
31.641
Time (s)
Fig. 9.11. Velocity variations during two-stage valve closure
function of valve position during closure. This information allows the

manufacturer to design an appropriate lever and weight which is able
to overcome resistance and allow the valve door to close within the
design period.
140
Base of
120
draw-off well
100
Valve 10% open Valve shut

Head (m)
80
60
30 s emergency valve closure
40 Upstream of butterfly valve
20
0
0.02
1.00
1.98
2.96
3.94
4.92
5.90
6.88
7.86
8.84
9.82
10.80
11.78
12.76
13.74
14.72
15.70
16.68
17.66
18.64
19.62
20.60
21.58
22.56
23.54
24.52
25.50
26.48
27.46
28.44
29.42
Time (s)
Fig. 9.12. Head variations for two-stage valve closure
116
Valve operations
9.4 Submerged discharge valve

It is also possible to improve the relationship between velocity and valve
position by employing a more sophisticated type of valve. One such is
the submerge discharge valve described by Miller (1969) (Fig. 9.13).
The principle of operation is that a ported sleeve, attached to a shaft
and actuator, is moved vertically to control flow passing into a sump
Actuator
Water surface
Spillway
Flowmeter
Inlet bend
Discharge sump
Vertical
downpipe
Corner fillets
Spindle
Valve outlet Moving sleeve
Baseplate
Fig. 9.13. Submerged discharge valve
117
where the energy of flow is dissipated. The port shape can be modified
to suit the circumstances both of the system in which it is installed and
also the range of flows over which the valve is to exercise control. The
port shapes can be altered over time to suit any changes in flow rate as a
scheme is developed. For example, in the early years of a project, flow
may be modest to suit prevailing demands while in later years flow
may have to be increased to meet anticipated increase in demand.
9.5 In-line valves

Two basic purposes can be identified for in-line valves. The most common
is the isolating valve which is normally fully open and remains so except
during periods of maintenance or possibly when an emergency has
occurred which requires the valve to be closed. The second valve applica-
tion is where an actuated valve is used to control and shut off flow within a
pipeline system but where a terminal valve would be impractical. The
method of analysis is the same for each category.
9.5.1 Isolating valves

While hydraulic transient analyses are commonly associated with
operation of terminal discharge valves for instance, some attention
should be paid to the action of isolating valves installed at intervals
on longer pipelines and also on branch pipelines. These valves are
included for the primary purpose of allowing sectional maintenance
without dewatering the entire system. Should an emergency arise,
such as a pipe burst, flow within the upstream pipeline will accelerate
towards the burst site. Isolating valves in the network adjacent to the
burst and in particular on the supply side of the system, must be able
to close safely and without causing further bursts.
Sluice valves and more recently butterfly valves are considered appro-
priate for this application. The butterfly valve is easier to operate than a
comparable sluice valve under similar conditions. For ease of manual
operation, handwheel diameter d has been given by Bartlett (1978) as:
p
d¼6 D
with D being pipe diameter. The practical continuous handwheel opera-
tion rate is around 20 rpm. Closing a valve against an unbalanced head
may require spur gearing. In one example, McCrone (undated) quotes
3 :1 gearing to close a sluice valve against a differential head of
40 mWG. The total number of turns till closure is 144, giving a 7.2 min
118
Valve operations
Compression wave travelling upstream

when isolating valve is closing
Steady flow hydraulic

gradient pre-burst
Gradient steepens
post-burst
Isolating valve Gradient flattens/reverses post-burst
Isolating valve
Initial flow direction

Initial pipe burst
Possible pipe burst
due to high pressure Possible pipe collapse due to
after valve closure low pressure after valve closure
Rarefaction wave travelling downstream

when isolating valve is closing
Fig. 9.14. Emergency closure of an in-line valve
closure interval. The butterfly valve by contrast has a standard 70 :1 worm

gear needing only 17 12 turns to close through 908. Closure time for the
butterfly valve is only 0.87 min. Surges produced by these valves were
93 m for the sluice valve and 135 m for the butterfly valve. These repre-
sent pressures 232% and 337% higher than the allowable maximum.
The effect of a too rapid isolating valve closure may be to produce
additional pipe bursts upstream due to the allowable peak pressure
having been exceeded. Also on the downstream side of the valve,
pressure will fall, potentially causing vacuum pressures which may
cause a pipe to buckle and collapse (Fig. 9.14).
9.5.2 Actuated valve

Figure 9.15 shows a water transfer system in which flow which has
previously been pumped from a reservoir to a break pressure chamber
(BPC) travels under gravity to a connection with a pumping main.
An in-line valve has been included upstream of the connection to
the second main. Flow passing through the in-line valve enters the
second main and continues to flow under gravity to augment the
volume stored in a second reservoir. The in-line valve is of a needle
119
Ground level = 10.8 mAOD

C.L. = 8.75 mAOD
Isolating valve Needle valve

Flowmeter Flow direction
Inlet main
Downstream main Floor level = 8.15 mAOD

Thrust block
M
BPC
Receiving
reservoir Flow direction
M
Flow direction
Valve Gravity transfer main

chamber
Downstream main
M
Suction tank
PS (idle)
Fig. 9.15. In-line needle valve
180
160
140
Head (mAOD)
120
100
80
60 Head upstream of needle valve –

1.8 min closure at a flow rate of 16 Mld
0 1 2 3
Time (min)
Fig. 9.16. Head upstream of needle valve for 1.8 min closure
120
Valve operations
45
42
39
Head (mAOD)
36
33
30
Head downstream of needle valve –
1.8 min closure at a flow rate of 16 Mld
0 1 2 3
Time (min)
Fig. 9.17. Head downstream of needle valve for 1.8 min closure
type whose head-loss characteristic is shown in Fig. 9.2 and which is
installed in a chamber as shown in Fig. 9.15.
When the valve is closed over a time of 110 s, head upstream of the
valve increases smoothly to a maximum as depicted in Fig. 9.16. After
wave reflection from the BPC, head falls to produce an irregular
oscillation. On the downstream side of the valve, head falls during
valve closure (Fig. 9.17). After valve closure and wave reflection, head
largely recovers and a fairly regular oscillation is developed. The timescale
of oscillations shown in Figs 9.16 and 9.17 are indicative of the respective
lengths of pipeline upstream and downstream of the closed in-valve.
Overall variations in head through the system can be seen in Fig. 9.18.
175
BPC
150
Envelope curves of max. and
125 min. head for 1.8 min closure
Head (mAOD)
Flow rate = 16 Mld

100
75
Downstream
reservoir
50
Needle valve chamber
25
Pipeline profile
0 1 2 3 4 5 6 7 8 9 10 11 12 13 14
Chainage (km)
Fig. 9.18. Envelope curves for 1.8 min valve closure
121
The much lower amplitude of oscillation downstream of the valve is

a function of pipeline length and diameter. The longer the pipeline
the greater will be the amplitude but the larger the pipe cross-section
the lower will be the velocity of flow and amplitude of surging
developed.
9.6 Control of transient pressures and estimation of valve

operating time
Essentially, control of hydraulic transient pressures using a valve which
is under operator control is a matter of limiting rates of velocity change.
For a given size and pattern of valve this requires determination of an
appropriate time of closure/opening so as to avoid unacceptable over-
pressure or underpressure.
It has been demonstrated how the retardation of flow in a pipeline
during valve closure is highly non-linear on account of the head loss
relationship of valves. Much of the flow deceleration and the peak
pressure developed will generally occur during this period of high
deceleration. It is always useful for the engineer to be able to make
rapid preliminary estimates of parameters and the necessary valve
closure time is just one of these.
The critical period of a pipeline is 2L=a and any valve which operates
to open or close in a time 2L=a will create an inertial pressure head
rise ¼ a V=g where V is the change in velocity at the valve.
This simple expression ignores the effect of attenuation.
When it comes to valve movements which occur in a time >2L=a
then the effects of wave reflection become significant with a reduced
peak pressure change. Some commentators have introduced a rule of
thumb which separates valve stroke times into rapid or slow movements
with stroke times >20L=a being defined as slow with elastic wave
effects being relatively unimportant.
Provided that valve movement is sufficiently slow that elastic wave
effects are not too important then an initial estimate of valve closure/
opening time can be made using the equation obtained from rigid
column theory. Neglecting pipeline resistance then, inertial head rise
during closure is given by:
Hi ¼ L=g dV=dt ð2:2Þ
If an assumption is made regarding closure time and maximum dV=dt
then Hi can be calculated. If head rise is excessive a more prolonged
closure time is required.
122
Valve operations
Table 9.1. Effective valve closure times
Valve type Effective time as a percentage

of actual valve closure time
Butterfly valve 21
Gate valve 39
Cone valve 49
Globe valve 52
For a slow valve closure Fox (1977) developed an equation of the

form:
p
Hmax =Ho ¼ 1 þ 2K 2 ð1 þ f1 þ 1=K 2 gÞ ð9:1Þ
where K ¼ Vo L=ð2gHo TÞ, Hmax is the maximum head, T is the closure
time, Ho is the upstream reservoir level relative to discharge level, Vo is
the initial steady velocity of flow, L is the pipeline length and g is
acceleration due to gravity. A linear reduction of effective valve area
over the stroke time was also assumed.
An alternative approach was described in an ASCE (1975) publica-
tion. This method can be applied to both rapid and slow closure times
and uses the concept of an effective valve closure time. The effective
valve closure time is defined as the time to reduce velocity linearly to
zero which produces the same transient pressure rise as the actual
valve closure time. The actual valve closure will generally produce a
non-linear velocity variation over the closure interval. Each valve
pattern has a different effective closure time as indicated in Table 9.1.
In the case of a gate valve for instance, an actual valve closure time of
100 s will impose a maximum deceleration dV=dt equal to a uniform
rate of deceleration over 39 s. It will be noted that this approach
becomes less viable when resistance effects are high.
In the absence of pipeline resistance the instantaneous head rise
Hmax produced when flow is decelerated in time < 2L=a is aVo =g in
Table 9.2. The ratio of head rise Hi during the closure in effective
time Teff , to the instantaneous head rise Hmax , is expressed in units
of 2L=a for different values of a pipeline constant aVo =ð2gHo Þ where
Ho is the initial steady flow head at the valve.
Suppose a gate valve is installed at the downstream end of a pipeline
of length 5000 m and initial velocity is 1.0 m/s. Let initial head Ho
be 20 m and wavespeed a ¼ 1000 m/s. Then the pipeline constant ¼
2.548 and critical period ð2L=aÞ ¼ 10:0 s. Instantaneous head
123
Table 9.2. Hi =Hmax as a function of pipeline constant and Teff
aVo =ð2gHo Þ Teff ¼ 5ð2L=aÞ Teff ¼ 10ð2L=aÞ Teff ¼ 20ð2L=aÞ

(seconds) (seconds) (seconds)
0.5—5.0 0.16 0.065 0.028

10.0 0.25 0.082 0.032
40.0 0.56 0.210 0.060
rise ¼ 1000ð1:0Þ=9:81 ¼ 101:937 mWG. If allowable head rise Hi is

chosen as say 16 m then Hi =Hmax 0:16 and Teff ¼ 5:0 10:0 ¼ 50 s.
Since Teff is 39% of the closure time for the actual valve then the
required valve closure time is 50=0:39 ¼ 128 s.
124
10
Pumps
Pumps are used to move fluids from point to point where insufficient
piezometric head is available to permit gravity flow. Pumping equip-
ment comes in a wide variety of forms to suit a range of different
functions. The potential for pressure transients to be initiated by each
type of pump is dependent upon the characteristics of the pump. The
majority of pumpsets in service initiate hydraulic transient events
when starting or stopping while other types of pump are themselves
incapable of initiating such events. Some designs of pumping plant
create transient behaviour as a part of their operation and in one
case the pump only functions by creating a hydraulic transient and
using its properties to lift water. In any discussion of hydraulic transients
it is therefore worthwhile to spend some time considering the types of
pump available and their characteristics as these will have a bearing
upon the nature of transient events developed and also on the types
of surge protection equipment appropriate in each instance.
10.1 Types of pump

The majority of pumps used in the water and sewerage industries can be
classified in a number of ways but from a transient standpoint pumps
may be grouped as discussed below.
10.1.1 Pumps which produce transient behaviour only when

changing their mode of operation — that is, starting, stopping
or changing speed
This category contains the majority of pumps used in water and sewerage
and includes the large family of centrifugal pumps or turbine pumps and
125
such positive displacement types as screw pumps where water is trapped in

spaces between the rotating element and the pump casing and is forced
through the pump to the discharge side.
10.1.2 Pumps which generate surge effects as a by-product of their

operation
This group includes reciprocating pumps in which a piston moves in
alternating directions within a close-fitting cylinder. While moving in
one direction the piston produces a partial vacuum into which water
flows through an intake port. As piston direction reverses water is
pressurised and forced out via an outlet port. The discharge from this
type of pump will tend to be pulsating to some extent. This second
group also includes the pneumatic ejector in which small sewage
flows are periodically discharged into a rising main using compressed
air and producing a transient event in that main.
10.1.3 Pumps which produce transient events in order to fulfil their

function
The hydraulic ram is an example of this third category. In this device
the flow of liquid in a supply pipe is arrested quickly by closing a
waste or impulse valve and producing a transient pressure rise upstream
of this valve. A delivery valve opens and the flow enters the air chamber
under the increased pressure. The increase in pressure allows flow to be
lifted to a header tank. Once the flow in the supply pipe starts to
reverse, the delivery valve shuts and the waste valve reopens to allow
a new cycle of operation to begin.
10.1.4 Pumps which do not by themselves produce surge effects

This group includes devices such as the jet pump where a supply of
liquid under pressure is forced through a nozzle or jet to entrain the
liquid to be pumped. An exchange of momentum occurs between the
high-velocity jet of ‘motive’ liquid and the low-velocity liquid to be
pumped. The combined flow exits via a diffuser.
10.2 Turbine pumps

Turbine pumps are characterised by one or more impellers equipped
with vanes, which rotate in a pump casing. These pumps are called
radial flow, mixed flow or axial flow pumps depending upon the
126
Pumps
Ns = 20–35 Ns = 35–60
Radial Radial
Ns = 60–90 Ns = 90–120
Francis/semi-axial/diagonal Francis/semi-axial or diagonal
Ns = 120–160 Ns = 160–300
Mixed flow Axial/propeller
Fig. 10.1. Impeller types
primary direction of flow through the impellers. Alternative terms also

used are centrifugal for radial pumps and propeller for axial pumps. The
mixed-flow types are intermediate forms between the purely radial and
axial types and are also called semi-axial, Francis or in some cases
diagonal pumps. Impeller forms for each type are illustrated in Fig. 10.1.
Since their use is so widespread, it is not surprising to find that the
majority of pressure transient analyses involves these pumps.
Turbine pumps constitute the largest group of pumps in use nowadays
for transfer of water and sewage. These will be primarily electrically driven
but in some circumstances diesel or other engines may be used.
10.2.1 Centrifugal or radial flow pumps

In radial flow pumps, water enters at the centre of the casing and flows
radially outwards being accelerated by an impeller rotating with angular
127
velocity ! rad/s. Water leaving at the periphery of the impeller is slowed

by the casing and a proportion of this energy is converted to potential
energy or pressure. This conversion may be achieved by the shape of
the casing itself as in a ‘volute’ pump or by internal diffusers or guide
vanes.
10.2.2 Mixed or semi-axial flow pumps

Pressure head is developed partly by centrifugal action and partly
through lifting action of vanes on the liquid. Flow enters axially and
discharge is through both centrifugal and radial motion. This category
of unit is often found within installations operating at low head and
requiring high-volume output.
10.2.3 Axial flow or propeller pumps

Pumping head is obtained principally through lifting action of vanes on
liquid. A single inlet impeller is used and flow enters axially and
discharges almost axially. Diffusion vanes may be installed on the exit
line to reduce swirl produced by the rotating propeller.
10.3 Turbine pump performance curves

In centrifugal pumps the pressure or the head Hp developed across the
pump is proportional to the kinetic energy of flow at exit from the impeller
V 2 =ð2gÞ and velocity V is proportional to the speed of rotation ! and
diameter D. It follows that for a family of pumps of varying diameter:
Hp ! 2 D 2 and Q !D3 ð10:1Þ
Output power Pr ¼ QHp , where ¼ g is the specific weight of
liquid flowing, is therefore given by:
Pr !D3 !2 D2 or Pr !3 D5 ð10:2Þ
If speed, discharge and head developed are known for a single pump
then
Q=! ¼ constant; Hp =ð!2 D2 Þ ¼ constant and Q=ð!D3 Þ ¼ constant
Eliminating D from the above relationships:
p
! ðQÞ=H3=4p ¼ constant ¼ !s ð10:3aÞ
where !s is called the specific speed.
128
Pumps
100
90
Q > 631 litre/s
Q>
= 631 litre/s
80
Q = 315 litre/s Q = 63.1 litre/s
Q = 31.5 litre/s
Efficiency (%)
Q = 12.6 litre/s
70
Q = 6.3 litre/s
60
50
40
10 20 30 40 50 60 70 80 90100 200 300
Specific speed (rpm ÷(m3/s)/m3/4)
Fig. 10.2. Specific speed plotted against efficiency
Since specific speed is not dimensionless it will have different numer-

ical values depending upon the system of units employed. All identically
shaped pumps have the same specific speed irrespective of their size.
The values used in calculating the specific speed of a pump are those
at the point of maximum efficiency. Rotational speed may be defined
as N rev/min where N ¼ 60!=ð2Þ, Hp may be metres and Q m3 /s.
Specific speed Ns in rev/min is then:
p
Ns ¼ N ðQÞ=H3=4 p ð10:3bÞ
From experience it is usually possible to estimate the impeller shape
which will give best results for a given speed, flow rate and delivery
head. Figure 10.2 shows the relationship between pump p efficiency
and specific speed for Ns expressed in units rpm ðm3 =sÞ=m3=4 .
When flow is small and head is relatively high, specific speed is small,
favouring a radial type of impeller. For large flow rates and low head
conditions, specific speed is high and the axial type is most suitable.
When pumping clean water, treated effluent or sewage that has been
129
passed through fine screens, radial pumps can be used with efficiencies
reaching 90%.
Other considerations other than pumping efficiency can be impor-
tant, for example in sewage pumping. When required to pass solid
material, the impeller shape can be modified to avoid choking and/or
to disintegrate matter present in the sewage. Some efficiency is
sacrificed and such pumps may have peak efficiencies in the 50—60%
range. Sewage pumps will usually operate at relatively low speed,
typically not greater than 960 rpm. This compares with speed of a
clean water pump which will usually be 1460 rpm and upwards. An
unchokable pump may be classified according to the size of solid
object which it can pass, with the normal maximum being specified
as 100 mm. For very small flows, such as from isolated properties, it is
not practical to make a centrifugal pump capable of handling the
H, P H, P
H–Q H–Q
P–Q
P–Q
Q Q
(a) (b)
H, P H, P
H–Q
H–Q
P–Q
P–Q
Q Q
(c) (d)
Fig. 10.3. Performance curve forms: (a) unstable, overloading; (b) stable, non-
overloading; (c) high shut-off head, flat power curve; (d) high shut-off head, high
shut-off power
130
Pumps
small volumes while retaining the ability to pass 100 mm diameter

solids. Other forms of pumping equipment must therefore be used.
For a single member of that family which is part of an hydraulic
transient analysis, diameter D is constant and so the relationships
10.1, 10.2 and 10.3 can be reduced to:
Q !; Hp !2 and Pr !3
or in terms of speed N expressed in rpm:
Hp N2 ; Q N and Pr N3
Behaviour of a typical centrifugal pump is usually described by perfor-
mance curves relating Hp , Q, Pr and N. The shape of the curves is
related to the form of impeller and specific speed. Typical shapes of
performance curves are illustrated in Fig. 10.3. Where it is necessary
to deliver a range of flows with only a relatively modest variation in
head then a flat curve relating Hp and Q is more suitable. If pumping
head varies considerably and it is desired to deliver an almost steady
flow then a steep Hp —Q curve is more attractive.
When a pump is starting/stopping, its speed varies continuously and a
family of performance curves is produced covering the speed range
(Fig. 10.4). Similar points on different curves will display the same
relationship between the variables so that plotting Hp =N2 against Q=N
Hp
Hp–Q curve at design speed
Duty point
Speed N reducing
Hp–Q curves at reduced speed
Fig. 10.4. H—Q performance curves for changing speed N
131
Hp/N 2
Q/N
2
Fig. 10.5. Hp /N against Q/N curve
then this family of curves can be represented by a single curve. Likewise

plotting Pr=N3 against Q=N produces a single curve (Fig. 10.5). As an
alternative to plotting power Pr torque, T ¼ Pr=! can be plotted.
Manufacturers will usually test pumps within the normal zone of opera-
tion — that is, þve flow and þve head — down as far as ‘runout’. This
leaves a considerable area of operation which will not commonly be
measured. This includes zones of þve flow and ve head, ve flow
and ve head and ve flow and þve head. To undertake hydraulic tran-
sient studies it is usually necessary to have information on these other
zones of operation. Some data are available from manufacturers covering
the complete operational characteristics of pumps. Donsky (1961) has
presented data for centrifugal pumps having different specific speed !s.
Such information can be used to synthesise performance curves for the
majority of pumps where such complete information is lacking.
10.4 Including turbine pumps in hydraulic transient

analyses
When in steady operation:
power input ¼ internal losses þ power output
When a pump is starting or stopping:
power input ¼ internal losses þ power output þ dE=dt
where E is the kinetic energy of the rotating elements comprising the
pumpset. Energy E can be written as:
E ¼ 12 I!2 ð10:4Þ
132
Pumps
where dE=dt ¼ power ðPrÞ is the rate at which energy changes with
time, I is the pumpset ‘moment of inertia’ and ! is the angular velocity
of rotation. Differentiating with respect to time then:
Pr ¼ 12 I d!2 =dt ¼ I! d!=dt ð10:5aÞ
Equation (10.5) can be written in terms of torque (T) where torque is
Pr=! so that:
T ¼ I d!=dt ð10:5bÞ
When the pump is accelerating, say when being started, d!=dt ¼ þve,
and when the pump is decelerating, as after being tripped,
d!=dt ¼ ve. After being tripped power input becomes zero. (Note:
for anyone unfamiliar with the term ‘moment of inertia’, a brief descrip-
tion of this parameter and its determination has been included in the
appendix to Chapter 11.)
The Hp =N2 —Q=N and power P=N3 or torque T=N2 —Q=N curves
can be digitised to yield either linear or parabolic segments for each
zone of pump operation. Using the head Hp , speed N and flow Q
relationship for the normal zone of operation as an example, Fig. 10.6
illustrates how a series of overlapping parabolic segments can be used
to represent the performance curves.
Angular acceleration d!=dt and torque T are related through the
moment of inertia I using the equation:
T ¼ I d!=dt ð10:5bÞ
For a parabolic approximation to a segment of curve:
Hp =N2 ¼ aH ðV=NÞ2 þ bH V=N þ cH ð10:6aÞ
T=N2 ¼ aT ðV=NÞ2 þ bT V=N þ cT ð10:7Þ
where aH , bH , cH and aT , bT , cT are the coefficients of the parabolic
segments relating to Hp , N and V and T, N and V. These coefficients
can be pre-computed before any simulation of pressure transients
commences, with the values stored and used during all studies involving
the particular performance curves.
At the start of any time increment, the velocity V or flow rate Q at the
pump and the speed of the pump N will be known. The value of Q=N can
be used to select the appropriate parabolic segment of curve and hence
the coefficients aT , bT and cT of the parabola. Torque T can then be
found at the start of the time increment. Speed change ! over the
time step t can then be found using:
! ¼ tT=I or ! ¼ !o þ tT=I ð10:5cÞ
133
Parabola centred
Hp/N 2 on I – 1
Limits of use for parabola

centred on point ‘I’
I–1
(Hp/N 2)I–1
I
(Hp/N 2)I Parabola centred
on I + 1
I+1
(Hp/N 2)I+1
Q/N
(Q/N)I–1 (Q/N )I (Q/N)I+1
Current (Q/N) value
Fig. 10.6. Digitisation of a curve
using the revised speed ! and initial velocity, the coefficients aH , bH and
cH of the parabola defining Hp can be obtained and hence the value of
pumping head at the close of the time increment.
Total moment of inertia I is made up of the individual inertias for the
motor, shafting, pump impeller and entrained liquid. Values of these
components can usually be obtained from pump and motor manufacturers
when a pumpset has been selected. For purposes of preliminary studies
before a particular pump supplier has been selected it may be necessary
to obtain a preliminary estimate of moment of inertia. Linton (1972)
has compiled data for both pump and electric motor inertias and presented
these in a graphical form covering a wide range of flow, head and speed.
These curves allow initial estimates of inertia to be obtained.
10.4.1 Transfer pump

For a transfer pump at the end of a pipeline, Fig. 10.7, then:
Hp ¼ H Hr or ¼ aH V2 þ bH VN þ cH N2 ð10:6bÞ
134
Pumps
Hp Piezometric line or
hydraulic gradient
Hr
M
H
Rising main
Tank or reservoir V +
Pump
Horizontal datum
J± = V ± g/aH
Dt
C± characteristic
Dx
Fig. 10.7. Transfer pump representation
Also, from the characteristic path arriving at the pump at the end of
the time increment, the quasi-invariant value J from Chapter 4,
gives the relationship just downstream of the pump:
V g=aH ¼ J or H ¼ ð J VÞ=ðg=aÞ
Substituting for H then:
ðÞa=gðJ VÞ Hr ¼ aH V 2 þ bH VN þ cH N2
or, rearranging:
aH V 2 þ fbH N þ ðÞa=ggV þ fcH N2 þ Hr ðÞa=g Jg ¼ 0
ð10:8Þ
This quadratic equation (10.8) can be solved for V. The corresponding
value of H at the end of the time increment can then be obtained
from:
H ¼ ð J VÞ=ðg=aÞ
135
The process can be repeated as required using the new assessments of

V and ! to improve the values of V, ! and H downstream of the
pump at the close of the time increment.
10.4.2 Booster pump

For a booster pump, where significant lengths of pipe exist both
upstream and downstream of the pump, hydraulic transients can be
important on both sides of the pumping station.
Considering Fig. 10.8, for the characteristics upstream and downstream
of the pump, the quasi-invarient values Jþ and J of Chapter 4 are
Hp
Piezometric line or
hydraulic gradient
Hd
Booster pump
Delivery main
Suction main Hu
V +
+
Horizontal datum
J+ = V + g/aHu J– = V – g/aHd
Dt
Dxu Dxd
Fig. 10.8. Booster pump representation
136
Pumps
obtained, giving the relationships:

V þ g=au Hu ¼ Jþ and V g=ad Hd ¼ J
or
Hu ¼ ð Jþ VÞ=ðg=au Þ and Hd ¼ ðV JÞ=ðg=ad Þ
The equations describing the digitised T=N2 —Q=N and Hp =N2 —
Q=N relationships are also available so that:
Hp ¼ Hd Hu ¼ aH V2 þ bH VN þ cH N2 ð10:6cÞ
or
ðV JÞ=ðg=ad Þ ð Jþ VÞ=ðg=au Þ ¼ aH V 2 þ bH VN þ cH N2
and rearranging:
aH V 2 þ fbH N gð1=ad þ 1=au ÞgV þ fcH N2 þ J þ Jþg ¼ 0
ð10:9Þ
This quadratic equation (10.9) can be solved for V and the corresponding
values of head upstream and downstream of the pump obtained by
substituting V in the equations Hu ¼ ð Jþ VÞðg=au Þ and Hd ¼
ðV JÞ=ðg=ad Þ.
10.4.3 Other pumping station and pipeline configurations

The arrangements discussed in sections 10.4.1 and 10.4.2 are probably
the more common types but there are other configurations which have
to be addressed from time to time. Consider the pumping station and
tunnel complex shown in Fig. 10.9. An intake structure is sited
upstream of a conurbation, on a river prone to flooding. When river
stage exceeds weir crest level, flow enters the upstream end of a
5.3 km long tunnel. At the downstream end of the tunnel a vertical
shaft allows flow to enter the pumping station where it is pumped
into the river downstream of the flood risk area. Four large concrete
volute pumps are used. No pressure waves from the tunnel impinge
directly on the pumps and the tunnel system lies entirely upstream of
the pumping station. When pumps are tripped, flow in the tunnel will
continue and the water level in the station upstream of the pumps
will rise and the pumps will turbine. It may be necessary to provide a
bypass arrangement to allow part of the flow to escape without
passing through the pumps.
137
Screens
Upstream river
Intake weir
Bellmouth inlet
Tunnel length 5.3 km
Design discharge = 60 m3/s with 4 pumps running
Flap gates
Penstocks
4 off concrete
volute pumps
Tunnel length
5.3 km
Fig. 10.9. Pumping system with long suction main
The same equation for the pumps is used but with the order of
upstream and downstream head reversed, that is:
Hd=s Hu=s ¼ aH V 2 þ bH VN þ cH N2
Other equations required to complete the model of this pumping

station are the conservation of volume upstream of the pumps together
138
Pumps
with the equation of the quasi-invarient along the characteristic at the

downstream end of the tunnel.
dHu=s =dt As ¼ Qin Qp Qbypass
Qin ¼ Vin At and Vin þ g=aHu=s ¼ Jþ
Qp ¼ VA
where Qin is inflow from the upstream tunnel, At is the tunnel cross-
sectional area and As is the free-surface area upstream of the pumps;
Hd=s is the river level downstream and Hu=s is the level in the suction
chamber. It is probably sufficient in most cases to calculate Hu=s as a
tank with outflow given by the initial discharge through the pumps
and bypass if any. The change in level over a time increment is obtained
using the single quasi-invariant equation from the tunnel.
10.4.4 Station losses

Expressions for head loss in pumping station fittings can be included in
the equations for head difference across a pump by representing the
head loss as:
X
H ¼ KL V 2 =ð2gÞVo =jVo j ð10:10Þ
The equation for head difference across the pump (and its fittings)
becomes:
X
Hp ¼ faH þ KL =ð2gÞVo =jVo jgV 2 þ bH VN þ cH N2 ð10:6dÞ
10.5 System curves and pump duty

When a pump is selected for a particular duty it will be required to
deliver the design discharge. Assessment of system conditions will
yield a system curve or curves. Normally the range in static lift together
with the anticipated variation of system resistance and pumping station
losses will be taken into consideration.
The best analysis of system resistance may be no better than 95%
accurate. The manufacturer and designer may both increase pumping
head and flow to ensure satisfactory service. For smaller pumps a 10%
allowance may be applicable and with a 5% margin in the case of
larger pumpsets. To ensure that the pump will be able to deliver the
required flow rate and head, a conservative pumping head will usually
be adopted.
139
The highest system curve will embody maximum static head and
maximum system resistance representing a mature main. In carrying
out surge analyses, usually the amplitude of pressure transient is a
function of flow rate and so computer simulation should also employ
the minimum assessment of system resistance to simulate clean main
conditions and also minimum static lift. The pump will deliver a
greater discharge when mains are new and static is low than when
mains are mature and static is high. Variation of system resistance is
particularly important in raw water and sewage lines. Resistance
changes will usually be less significant for treated water mains where
a pipe lining is included.
10.6 Turbine pump start

Pump start can be achieved by a number of methods: direct start, Star/
Delta, transformer and various types of variable speed start. These are
now discussed in turn.
10.6.1 Direct start

Direct start is the simplest approach with the pump running up to its
design speed quickly. There is a substantial electrical current drawn
from the electrical supply network during the starting operation and
the electricity system must be able to handle large starting currents.
Starting results in a voltage drop which will affect both the pump
motor and other equipment connected to the supply grid. Electricity
suppliers may impose restrictions on the size of motor which can be
started in this way.
10.6.2 Star/Delta and transformer starting

The Star/Delta starting arrangement employs two different connection
configurations within the motor. In Star mode, for example, a three-
phase motor intended to receive 380 V will obtain only 220 V with
only one-third of the stator windings being used. In consequence the
motor only develops a small part of its normal starting torque with
the effect of prolonging start time. When the set has achieved
around 80% of design speed an automatic change in connections to
Delta mode is made allowing the motor to receive its full voltage.
Start time should not be too long otherwise heat build-up may
damage the motor. Star/Delta may be unsuitable for large pump
140
Pumps
motors as in start mode the motor may not develop sufficient torque to
allow the pump to accelerate. An approximate limit of 30 kw has been
quoted as a maximum for Start/Delta starting.
Change-over from Star to Delta can cause problems. Timing is very
important and also the method of switching. Open switching completely
breaks the circuit during transition. This break, and the subsequent
switch to Delta mode can produce high transient currents and large
negative torques. These currents may damage windings and the
torques can damage the shaft between motor and pump. A flexible
coupling will alleviate this effect.
Where Star/Delta is unsuitable, transformer starting may well be an
option although more costly. The supply voltage in the initial stage is
reduced to a percentage of mains supply voltage through use of a trans-
former. Typically three tappings are provided on the transformer to
allow for difference between actual and anticipated values. These
tappings may represent voltages in the range 60—80% of mains
voltage. As the motor approaches its maximum speed the supply is
switched onto mains voltage. Complicated switching arrangements
reduce transient currents during switching.
10.6.3 Variable speed or ‘soft’ start

A large number of alternative arrangements of variable speed drives are
available. These allow the engineer to achieve fully automatic control of
pumping in many instances. Synchronous motors, induction motors and
frequency converters all permit a more prolonged start time to be
achieved so that both pump start and normal shutdown surge effects
can be reduced from those that would be attained with a straight-
forward direct start. The pump can be gradually ‘ramped’ up and
down, with the changing speed producing corresponding adjustments
in discharge and delivery head.
10.7 Case studies of pump start
10.7.1 Simulation of direct start in solo pumping

The pump is attempting to deliver into a rising main filled with a static
column of liquid. Not surprisingly the pump will begin by operating
close to its shut-valve condition with the only flow developed being
that which can be accommodated within the pipeline due to compres-
sion of the water and expansion of the pipe cross-section under the
141
Carlops rising main – start of duty pump. Head at pumping station and mid-point of main
400
390
380
370
Head (mAOD)
360
350
340
330 Ch. –0 km
Ch. –3.4 km
320
0.055
2.255
4.455
6.655
8.855
11.055
13.255
15.455
17.655
19.855
22.055
24.255
26.455
28.655
30.855
33.055
35.255
37.455
39.655
41.855
44.055
46.255
48.455
50.655
52.855
55.055
57.255
59.455
61.655
63.855
Time (s)
Fig. 10.10. Head variations for direct start of solo pump
pressure which will be close to the pump’s shut-valve delivery head. Initi-
ally, head in the rising main will exceed the design head. Figure 10.10
shows this for a solo duty pump delivering into a single rising main
carrying treated water. After a delay while the startup compression
pressure wave travels along the main, points along the pipeline are
affected by the pump start and experience a head rise similar to that
at the pump. Figure 10.10 also shows head conditions at the mid-
point of the pipeline, chainage 3.4 km from the pumping station.
After time L=a the pressure wave reaches the downstream end of the
pipeline and a rarefaction wave reflection travels back along the main
towards the pumping station, bringing with it a relief of pressure.
Points further from the pumping station experience this relief of
pressure first, with a fall in head occurring at the mid-point of the pipe-
line. When the reflected pressure wave reaches the pumping station a
fall in head occurs to a lesser extent as the pump operating point
moves to a lower delivery head and higher flow. The pressure waves
continue to travel to and fro in the main, with each wave reflection
producing a fall in pump delivery head and corresponding increase in
flow. Effects of developing pipeline resistance as flow increases gradually
diminish the transient effect, with an almost steady flow occurring after
142
Pumps
Carlops rising main. Start of duty pump. Max. and min. head along main
400
350
300
Elevation (mAOD)
250
200
150
100
Invert level (i.l.) (mAD)
h (max.)
50
h (min.)
0
0
280
560
840
1120
1400
1680
1960
2240
2520
2800
3080
3360
3640
3920
4200
4480
4760
5040
5320
5600
5880
6160
6440
6720
7000
Chainage (m)
Fig. 10.11. Envelope curves for direct start of solo pump
1 min of operation. After the initial upsurge, pressure wave amplitudes

at points further along the pipeline are more noticeable than those at
the pumping station.
As far as overall head change is concerned, Fig. 10.11 shows that the
initial upsurge when the pump is operated creates an almost constant
surge amplitude along the main.
10.7.2 Direct start in multi-pump operation

In a multi-pump installation, the option of starting individual pumps in
sequence is available. A pipeline network is subject to pressure transients
from start of each pump and a sufficient time delay can be incorporated to
allow transient effects from starting of one pump to have decayed to a large
extent before the next pump is started. In this way start-up transient
pressures can be limited and also beneficial effects can accrue by reducing
maximum demand charges for the electrical supply network.
Figure 10.12 shows predicted variations of head in a sewage pumping
system comprising two parallel mains having diameter DN 450 and
DN 600. Two duty pumps are available and these were started in
sequence with a 30 s interval between the start of each pump. Direct
start was used for each pump. When the first pump is operated, head
rises steeply at the pumping station, to a short-lived peak before
falling back by a modest amount. This fall in head is produced by a
143
Sharjah PS No. 3. Start of two duty pumps with 30 s between starts

70
60
50
Head (mASL)
40
30
20
10 PS
Ch. 1525 m
0
0.173
3.114
6.055
8.996
11.937
14.878
17.819
20.760
23.701
26.642
29.583
32.524
35.465
38.406
41.347
44.288
47.229
50.170
53.111
56.052
58.993
61.934
64.875
67.816
70.757
73.698
76.639
79.580
82.521
85.462
Time (s)
Fig. 10.12. Sequenced direct start of two duty pumps
pressure wave reflection from the bifurcation into two mains down-
stream of which there is an overall increase in cross-sectional area.
Thereafter, head starts to rise more slowly. About halfway along the
mains, at chainage 1.525 km, head rises when the pressure wave
reaches this location (Fig. 10.12).
Following reflection of this start-up pressure wave from the outfall
end of the system, head falls within the mains. Those points closest
to the outfall experience the effects of pressure relief first, as at the
mid-point of the mains (Fig. 10.12), with the pumping station being
affected after about 21 s. At 30 s the second pump is operated while
pressure is relatively low at the pumping station, and a second upsurge
pressure wave is created. It will be noted that this second upsurge is
smaller than the first. This is common in multi-pump installations, with
successive pump starts having a progressively smaller effect.
Subsequent wave reflections progressively reduce system pressures
with near-steady conditions achieved after about 112 min of starting
the first pump. The difference in head between the pumping station
and the halfway point is indicative of pipeline resistance.
Development of flow in the two mains follows a stepped pattern as
shown in Fig. 10.13. Velocity in each main is initially similar but as
144
Pumps
Sharjah PS No. 3. Start of two duty pumps. Velocity variations

3.5
2.5
Velocity (m/s)
1.5
0.5
Ch. 0450
Ch. 0600
0
0.173
2.941
5.709
8.477
11.245
14.013
16.781
19.549
22.317
25.085
27.853
30.621
33.389
36.157
38.925
41.693
44.461
47.229
49.997
52.765
55.533
58.301
61.069
63.837
66.605
69.373
72.141
74.909
77.677
80.445
83.213
85.981
Time (s)
Fig. 10.13. Flow development for start of two pumps in sequence
Sharjah PS No. 3. Envelope curves for start of two duty pumps

70
60
50
40
Elevation (mASL)
30
i.l. (mASL)
20 h (max.)
h (min.)
10
0
0
122
244
366
488
610
732
854
976
1098
1220
1342
1464
1586
1708
1830
1952
2074
2196
2318
2440
2501
2623
2745
2867
2989
3111
3233
–10
Chainage (m)
Fig. 10.14. Envelope curves for direct start of two pumps in sequence
145
their different resistance characteristics start to influence matters, the

larger DN 600 main achieves a significantly higher velocity.
Maximum head along each main shows a gradual fall between the
pumping station and the outfall (Fig. 10.14).
10.8 Initial conditions of flow

In conducting pressure transient studies of pumping systems, as with
gravity pipelines it may simplify matters if the system is started from
rest. This allows static initial conditions to be specified, avoiding the
need to predetermine an initial steady flow condition and hydraulic
gradient. Start-up analyses can also be useful in several respects:
(a) Prediction of start-up transient pressures and time taken for these
transient pressures to decay in order to establish essentially steady
flow.
(b) Determining an appropriate time delay to be allowed between
successive pump starts in a multi-pump installation to avoid any
adverse interaction between surging produced by each pump.
(c) Providing a convenient means of establishing steady flow condi-
tions throughout the network.
Unless particular measures are included to create a ‘soft start’ then

the upsurge developed as a pump is operated will be dependent upon
the pump/motor and system characteristics — that is, dV=dt is not
easily regulated. Likewise, when a pump is tripped, only the energy
contained in the rotating elements 12 I!2 is available to sustain
pumping. In many instances the moment of inertia I is quite modest
and so the amount of stored energy is small, leading to relatively
rapid rates of deceleration and head change at the pumping station.
10.9 Pump failure or ‘trip’

As far as the pipeline system as a whole is concerned, it is important to
examine full station trip or station ‘blackout’ and with a range in system
resistance. These studies will include the maximum pump discharge
using all available duty pumps in an endeavour to find the ‘worst
case’. Other aspects of hydraulic transient behaviour, such as the
response of a check valve following a pump failure, may dictate that
other pump combinations be considered to determine the most critical
event. This aspect is discussed in Chapter 20.
146
Pumps
Pump trip is an important case for simulation. The initial pressure

drop downstream of a pumping station can produce unacceptably low
pressures but also a subsequent upsurge after flow has reversed is an
important event which requires to be predicted.
In a pumping station containing only a solo duty pump, trip of the
pump will be analysed over the range of suction levels.
In a multi-pump installation wherein several duty pumps may be oper-
ating together it is important to simulate the effect of different numbers of
duty pumps being tripped, again using the pertinent suction levels.
For a multi-pump installation the total amount of stored energy in
rotating elements 12 I!2 varies directly in proportion to the number of
pumps in operation, whereas the increase in flow rate produced by starting
of each additional pump diminishes. The available ‘stored’ energy in
rotating elements per unit of flow, increases as total flow increases.
Thus it is not always obvious which pumping circumstance will produce
the ‘worst case’ as far as the system as a whole is concerned.
When pumping head is relatively high, the deceleration of flow after
pumps are switched off takes place with minimum head downstream of
the pumps remaining above suction levels. In this case the check valves
on each operating pump delivery branch will shut as flow ceases and
starts to reverse. Consider the long water pipeline shown in Fig. 10.15.
Figure 10.16 shows the predicted head downstream of the pumping
station when both operating pumps are tripped simultaneously at
time ¼ 1 s, simulating a station ‘blackout’. No protection is included
and head falls by almost 200 m at the pumping station over 4 s. At
this time, check valves close on each pump delivery branch. The
head drop is approximately given by aVo =g, where Vo is the initial
steady flow velocity in the rising main. However, head was predicted
to continue falling gradually until around 22 s when the effects of
wave reflections start to influence conditions. This further reduction
in piezometric level is a consequence of pipeline resistance and is a
similar effect to ‘attenuation’ as described in Chapter 7. At 5 km the
effects of pump trip are experienced around 5 s after trip and the
decline in head follows the same pattern as at the pumping station.
The influence of wave reflection is noticeable at around 10 s. At
10 km the falling head is arrested by operation of nearby air valves.
Many pumping systems, particularly sewage schemes, operate at more
modest head and the response of the pumps can be influenced by the
prevailing level in a wet well.
Consider the variations of head shown in Fig. 10.17, downstream of
pumps and part way along sewage rising mains. After all three operating
147
Az Zour to Wafra Main. Trip of two pumps. Max. and min. head along main
400
i.l. (mPWD)
350
h (max.)
h (min.)
300
Head (mPWD)
250
200
150
100
50
0
0
1153.36
2367.56
3532.11
4751.58
5983.65
7155.67
8448.01
9605.88
10 823.28
11 979.81
13 198.89
14 418.39
15 634.59
16 858.59
18 029.24
19 328.61
20 559.99
21 843.30
23 005.95
24 300.18
25 532.26
26 754.46
28 037.77
29 262.87
30 552.27
31 841.67
33 075.07
34 372.03
35 607.23
Chainage (m)
Fig. 10.15. Envelope curves for long rising main after pump failure
pumps are tripped, delivery head falls steeply until discharge head is
below the suction well level. Under the action of the suction well
level, sewage continues to flow through the pump, keeping the check
valves partially opened. The head downstream of the pumps is largely
Az Zour to Wafra Rising Main. Trip of two duty pumps. Head at PS, ch. 5 km and ch. 10 km
250
D/s pumps
Ch. 5 km
200
Ch. 10 km
Head (mPWD)
150
100
50
0
0.056
1.344
2.632
3.920
5.208
6.496
7.784
9.072
10.360
11.648
12.936
14.224
15.512
16.800
18.088
19.376
20.664
21.952
23.240
24.528
25.816
27.104
28.392
29.680
30.968
32.256
33.544
34.832
36.120
37.408
38.696
Time (s)
Fig. 10.16. Head variations in a long rising main after pump failure
148
Pumps
Egaila, Kuwait. Simultaneous trip of three pumps. Head at PS and ch. 2.6 km
160
140
120
Head (mPWD)
100
80
60
40
Ch. –0 km
20
h –2.6 km
0
0.116
1.160
2.204
3.248
4.292
5.336
6.380
7.424
8.468
9.512
10.556
11.600
12.644
13.688
14.732
15.776
16.820
17.864
18.908
19.952
20.996
22.040
23.084
24.128
25.172
26.216
27.260
28.304
Time (s)
Fig. 10.17. Simultaneous trip of three pumps in a sewage scheme
stabilised, showing only a slow and small recovery till around 14 s when
head increases more rapidly. Part way along the mains at chainage
2.6 km the fall in head is similarly halted. Figure 10.18 shows the corre-
sponding variation of velocity at the start of rising mains. Following an
Egaila, Kuwait. Trip of three pumps. Velocity at start of mains
1.6
1.4
1.2
1.0
Velocity (m/s)
0.8
0.6
0.4
0.2
Ch. 0.0 m
0 Time (s)
0.116
0.580
1.044
1.508
1.972
2.436
2.900
3.364
3.828
4.292
4.756
5.220
5.684
6.148
6.612
7.076
7.540
8.004
8.468
8.932
9.396
9.860
10.324
10.788
11.252
11.716
12.180
12.644
13.108
13.572
14.036
Time (s)
Fig. 10.18. Velocity changes after trip of three pumps
149
Egaila, Kuwait. Trip of three pumps. Max. and min. head along mains
160
i.l. (mPWD)
140
h (max.)
h (min.)
120
Elevation (mPWD)
100
80
60
40
20
0
0
200
400
600
800
1000
1100
1300
1500
1700
1900
2000
2200
2400
2600
2800
3000
3100
3300
3500
3700
3900
4000
4200
4400
4600
4800
4900
5100
Chainage (m)
Fig. 10.19. Envelope curves in a sewage rising main after pump failure
initial steep fall, velocity rises slightly before beginning a relatively

steady decline till around 14 s when the check valves shut. Envelope
curves in Fig. 10.19 show that this system is subject to large vacuum
pressures over most of its length and only the influence of the suction
level has prevented lower head at the pumping station.
10.10 Other pumps
10.10.1 Reciprocating pumps

Principles of operation of a positive displacement pump are relatively
simple. For each working cycle, revolution or stroke, a quantity of
fluid is enclosed and moved from pump intake to outlet. The flow
rate or volume enclosed per cycle is solely dependent upon the dimen-
sions of the displacement pump cavities. Attainable pressure head is
only dependent upon the mechanical strength of the pump and the
available driving power. A safety valve may be incorporated to limit
pressure increase. Figure 10.20 shows a schematic of a piston type of
pump. For many displacement pumps the delivered volume flow
varies during the course of a cycle of operation. The most extreme
example of output variation is that of a single-acting, single-cylinder
piston pump (Fig. 10.21(a)). Smoother flow can be achieved on both
suction and delivery lines by using a number of cylinders in parallel.
150
Pumps
Piston
Discharge
Suction
Fig. 10.20. Features of a reciprocating displacement pump
Time
(a)
Time
(b)
Time
(c)
Fig. 10.21. Flow against time relationships for piston pumps: (a) single cylinder
single-acting; (b) twin cylinder single-acting or single cylinder double-acting; (c)
three cylinder single-acting
151
P and Q
Flow Q against head Hp
Power P against head Hp
Hp
Fig. 10.22. Performance curves for displacement pumps. (Performance curves of

turbine-type pumps are not convenient due to steepness of H—Q and P—Q
curves)
Pump power requirements are correspondingly smoothed by this

approach. Figure 10.21(b) shows typical output from either a double-
acting single cylinder or single-acting twin cylinders. Figure 10.21(c)
shows flow from a three-cylinder single-acting unit. Built-in accumula-
tors (small air cushions) are often included on suction and delivery
branches as a means of smoothing flow. Behaviour of air vessels is
discussed in Chapter 12.
Using the turbine pump form of performance curves of head and
power against flow for these pumps produces almost vertical gradients
and for practical use it is more convenient to use curves of flow and
power as a function of head as illustrated in Fig. 10.22.
10.10.2 Pneumatic ejector

For sewage applications involving small flows and requiring the
capability to pass solids of up to say 100 mm diameter it is not practical
to make a centrifugal pump capable of meeting both of these criteria.
One solution is to use a pneumatic ejector, the main components of
which are shown in Fig. 10.23. A compressor delivers air to a receiver.
Sewage flows enter via gravity sewers to collect in ejector cylinders. A
float is located in each cylinder and when this float rises to a defined
level the air inlet valve to that cylinder opens to allow air to enter
thus pressurising the contents of the cylinder. When sufficient pressure
has developed to overcome pressure in the outlet pipeline system then a
non-return valve opens automatically to permit discharge of the ejector
152
Pumps
Air vent pipe
Air compressor
Compressed Compressed air
air receiver supply pipe
Discharge
pressure pipe
Inlet gravity sewer
Isolating valve
Check valve
Ejector cylinder
Check valve
Elevation
Check valve
Check valve
Isolating valve
Isolating valve
Inlet gravity sewer Discharge

pressure pipe
Ejector cylinder
Isolating valve
Isolating valve
Check valve
Check valve
Air exhaust pipe Air supply pipe

Plan
Fig. 10.23. Components of ejector station
153
cylinder contents to the downstream pipeline system. Twin cylinders

are often provided with one filling while the other is discharging.
Usually one compressed air system supplies both cylinders. Overall effi-
ciency will typically be in the range 20—50%. Ejector stations are
designed to handle flows from isolated premises or groups of houses at
inflow rates of between 1 litre/s to 10 litres/s at delivery heads of up to
a maximum of about 30 mWG.
Each operation of an ejector cylinder will produce a transient within
the downstream pipeline system. Ejectors are normally rated according
to the size of cylinder, for example 150 litres, 250 litres. Discharge
normally will be completed in around 60 s so that corresponding
average flow would be 2.5 litres/s or 4.2 litres/s. Actual discharge rate
may well exceed these values, possibly reaching twice the average, i.e.
5.0 litres/s or 8.4 litres/s, if the air supply is generous. In a DN 100
rising main this would produce a velocity change of 0.6—1.1 m/s and
a corresponding inertial pressure rise.
It is not uncommon to find several ejector stations teed into a rising
main which also acts as a collector pipeline for these stations. Each time
an ejector operates it will induce a pressure transient within the rising
main. The flow rate from the ejector will be dependent to some
extent upon the piezometric level in the rising main. Figure 10.24
illustrates this situation with flow from the pumping station into the
rising main.
Pressure waves travelling upstream and downstream

from ejector connection
dx/dt = V – a
DH
H
dx/dt = V + a
Steady flow piezometric line
Datum
Vu/s Vd/s +
Qe Collector/rising main
Pumping station
Ejector station connection
Fig. 10.24. Operation of an ejector station on a sewage rising main
154
Pumps
If the discharge from the ejector is Qe then with reference to Fig. 10.24
the equations required for solution at the ejector connection are:
from characteristics
Jþ ¼ Vu=s þ g=aH and J ¼ Vd=s g=aH
and from conservation of volume
Vu=s A þ Qe ¼ Vd=s A
eliminating Vu=s and Vd=s then,
H ¼ ð Jþ J þ Qe =AÞ=ð2g=aÞ ð10:11Þ
To obtain an idea of the magnitude of the inertial pressure rise caused
by an ejector, assume that there is no time for wave reflections to occur
prior to the maximum discharge from the ejector being achieved. Then
if Jþ and J remain at their initial steady flow values, the initial head
Ho at the ejector connection before discharge Qe occurs will be:
Ho ¼ ð Jþ JÞ=ð2g=aÞ
and when the maximum discharge from the ejector is Qemax , maximum
head will be:
Hmax ¼ Ho þ Qemax =ð2Ag=aÞ
or peak inertial head
H ¼ Qemax =ð2Ag=aÞ ð10:12Þ
Suppose Qemax ¼ 5 litres/s, D ¼ 300 mm (A ¼ 0:0707 m2 ) and a ¼
1000 m/s. The maximum inertial head increase will be:
H ¼ 0:005=ð2 0:0707 9:81=1000:0Þ ¼ 3:6 m
In a smaller main, say D ¼ 200 mm,
H ¼ 0:005=ð2 0:0314 9:81=1000:0Þ ¼ 8:1 m
The smallest sewage rising main may have diameter D ¼ 100 mm
giving: H ¼ 0:005=ð2 0:00785 9:81=1000:0Þ ¼ 32:4 m.
In a plastic rising main these inertia head values would be reduced
proportionately for the smaller wave speed in these pipes.
10.10.3 The hydraulic ram

This pump is an example of a device which produces a hydraulic
transient for the purpose of lifting water to a higher level. The
155
Header tank
Supply
reservoir
M
Air vessel
Riser
Non-return valve
Supply pipeline
Fast-closing valve
Fig. 10.25. Schematic of hydraulic ram
principle of operation may be illustrated with reference to Fig. 10.25.

The supply pipeline has to have sufficient length and diameter that a
worthwhile amount of kinetic energy is contained within the flow,
typical lengths being between 5 and 20 m. The air vessel is located
from 0.5 to 3 m below the water surface in the supply reservoir and
the difference in level between the supply reservoir and the waste
valve can be up to 6 m. If the waste valve is suddenly closed the
water column in the supply line is rapidly decelerated with an atten-
dant pressure rise. This rising pressure pushes upon the non-return
delivery valve, allowing flow to enter the air vessel and pressurise this
chamber, causing flow to pass through the riser pipe into the elevated
storage tank. Once flow in the supply line comes to rest and starts to
reverse, pressure below the delivery valve falls and this valve closes.
Falling pressure can also be used to reopen the waste valve allowing a
further cycle of operation to commence. Overall efficiency of the
process is around 50%.
The hydraulic ram has been described as an hydraulic transformer
converting low pressure into high pressure.
10.10.4 The jet pump

Jet pumps or injectors operate by transference of momentum from a
high-velocity jet of ‘motive’ fluid to a ‘pumped’ fluid medium within a
mixing tube (Fig. 10.26). A jet pump may be installed in a borehole
with a centrifugal pump at the surface providing the motive force for
the jet pump.
156
Pumps
H1
Piezometric levels
H3
H2
Qp Pumped fluid
2
Diffuser 3
Mixing tube
1
Qm Qp + Qm
Motive fluid
Nozzle
Fig. 10.26. Principles of a jet pump
The performance of this unit is described using two quantities. The

relationship between pumped mass flow and motive mass flow is
called the flow coefficient q, thus:
q ¼ p Qp =ðm Qm Þ
Or in the case where water comprises both motive and pumped media:
q ¼ Qp =Qm ð10:13Þ
the pressure or head relationship Z for a jet pump can be defined as:
Z ¼ ðH3 H2 Þ=ðH1 H3 Þ ð10:14Þ
Efficiency of the pump is then:
¼ qZ ¼ Qp ðH3 H2 Þ=½Qm ðH1 H3 Þ ð10:15Þ
Performance curves for a pump of this type may have the form shown
in Fig. 10.27. In any analysis there are five unknowns: H1 , H2 , H3 , Qm
and Qp . The equations available for solution at any time comprise three
characteristic equations, the head relationship and the flow coefficient
157
h, Z
Z–q
h–q
Fig. 10.27. Performance curves for a jet pump
relationship. These are sufficient to yield values of the dependent

variables after each time increment.
When the jet pump is without moving parts it will merely act to
partially transmit and reflect transients developed elsewhere in the
pipeline system and does not of itself initiate any pressure transient
effects. Some jet pumps are fitted with a control valve, usually in the
form of a needle valve, which can be used for discharge regulation.
Should the valve setting be adjusted then the jet pump will initiate a
transient event.
158
11
Flywheels
The concept of a reservoir of energy, however modest, being stored in

the rotating elements of a pumpset was established in the previous
chapter. The present section develops this idea as a means of surge
suppression by considering how the amount of energy, contained
within the rotating pump system, may be enhanced to improve the
ability to control rates of flow and pressure change at the pump after
it is tripped.
The rate of change of velocity at a pump determines the severity of
the hydraulic transient developed when the pump is started or
stopped. During start-up the changes in velocity will be controlled by
the mode of starting and to some extent any surge problems associated
with pump start can be designed out by an appropriate choice of starting
method. For instance, if rates of velocity increase produce unacceptable
surges then a flow control valve may be a possible remedy. This valve
can be used to regulate the way in which flow rate is allowed to
develop. During a ‘normal’ pump shutdown under either automatic or
operator control, the rate of deceleration of flow can again be
controlled. In the event of a power failure, on the other hand, the
pump speed and also pump discharge will reduce in a manner controlled
by the characteristics of the pumpset, liquid and pipeline system. The
only source of energy remaining to allow pumping to continue is that
stored within the rotating elements of pump, shafting and motor.
11.1 Moment of inertia

The stored energy E ¼ 12 I!2 in the rotating elements of a centrifugal
pump installation is dependent on the moment of inertia I ðkg:m2 Þ
for a given speed of rotation ! (rad/s). As the pump decelerates
159
following a pumping failure, energy is transferred to the water from this

store of energy. Increasing the moment of inertia is an obvious way in
which the amount of stored energy can be increased. After a pump is
tripped, this additional energy in the rotating elements can be used
to prolong pumping thus reducing rate of flow deceleration dV=dt
and with it the severity of pressure transient effects. This increase of
moment of inertia can be achieved by adding a flywheel to each
pump/motor shaft. Shearing of the shaft between flywheel and pump
impeller will render the flywheel ineffective.
11.2 Flywheels
The flywheel consists of one or more metal disks rotating on the same
shaft as the pump and motor and with the same speed. Figure 11.1
shows a typical installation for a vertical pump.
On the delivery side of a pumping station after pump failure, the rate
at which head decreases and flow decelerates is reduced by adding
additional inertia. In the case of a booster station, on the suction side
of the station, rate of head rise and rate of flow deceleration are also
Shaft to electric motor Upper bearing
Flywheel assembly
Lower bearing
Coupling
Shaft extending
downwards to pump
Fig. 11.1. Flywheel installation
160
Flywheels
reduced by adding a flywheel. After flow has ceased and the check valve
on each pump delivery branch has closed, the pump/flywheel arrange-
ment plays no further part in the transient event in a rising main. The
flywheel may have been added for the purpose of controlling either
minimum or maximum transient pressures in the main. Where peak
pressures are to be controlled then these will arise as a consequence of
deceleration of a reversed flow which has occurred subsequent to the
initial downsurge following pumping failure. The only way in which the
flywheel can limit such maximum pressures is by previously controlling
these initial minimum piezometric levels. The adverse hydraulic gradient
producing flow deceleration and accelerating flow in the reversed
direction in the main is thus limited. So it is only by limiting the
initial downsurge that the subsequent upsurge is controlled.
The flywheel forms an integral part of the pumpset assemblage and it
is speed change of the machine that initiates the transient event. By
adding the flywheel the rates of speed change are reduced and the
pressure transient is controlled at its source. The entire system benefits
from the flywheel solution. This contrasts with some other forms of
surge alleviation in which the hydraulic transient is allowed to
develop and its subsequent spread through the majority of a pumping
system is restricted by additional surge equipment. Reducing rates of
velocity change at the pump itself gives check valves time to respond
thus reducing potential for valve slam. The same applies during pump
start if the flywheel means that a slower acceleration of pump occurs.
The check valve will not then be thrown open.
This form of hydraulic transient suppression also has the advantage of
ease of maintenance.
Where a gearbox is included, relative speeds of pump and motor
require to be considered (Miller, 1978). Resultant inertia is expressed
in terms of pump speed so that
I ¼ Iimp þ Ishaft þ Imotor f!motor =!pump g2
11.3 Limitations on flywheel size

Sometimes the size of flywheel which can be added is limited by one or
more factors. The extent to which inertia can be increased is usually
determined by capability of the motor and starting equipment. Limita-
tions on the power of motor that can be operated using direct-start
means have already been mentioned in Chapter 10. To illustrate the
potential limitation of the flywheel solution, in one Middle East
161
sewage pumping scheme, transient analysis indicated that a flywheel

moment of inertia ¼ 150 kg.m2 was necessary to prevent vacuum
pressures in a glass reinforced plastic (GRP) main 1056 m long and
350 mm diameter. The pump motors already on order could accelerate
a total inertia only 12.7 kg.m2 . This limitation rendered the flywheel
option impractical. A change to a more complicated starting switch
arrangement would have been necessary.
The resulting size and weight increase following addition of a flywheel
requires careful study of station loading and structural vibration. Extra
bearings may be necessary, particularly in the case of vertical pumps.
Since the flywheel is usually installed between the motor and the
pump, should the impeller become uncoupled from its shaft the flywheel
will be rendered ineffectual. Again, if a pump blockage should occur the
flywheel will be unable to suppress the subsequent transient. A loss of
pump prime will render the flywheel ineffective. Other considerations
include available space within the pumping station and practicality of
fitting the flywheel to a particular pumpset arrangement. Other
penalties associated with a flywheel installation are power losses in
motor windage and an uprated starting arrangement. Finally, the
attitude of the client towards the flywheel as a means of transient
suppression may be an important factor.
Having decided to install a flywheel, calculations of the continued
flow after trip are important to ensure that sufficient reserve volume
remains within a wet well. Suppose a pump is tripped normally at its
stop level. The extra pumping that occurs with the flywheel added
will continue to draw liquid from the wet well at levels below
minimum operating level. The consequent reduction in well level
may start to draw air into the pump suction. This should be avoided.
11.4 Pipeline limitations

Manuel (1970) has indicated that the likely size of scheme which can be
protected against unacceptable pressure transients by means of a
flywheel is likely to be of the order 2 km in length and of modest
diameter. Fox (1977) also indicated a pipeline length of around 1—
2 km as being the sort of system which might be suitable for protection
using a flywheel. Suitability of a pumping system for a flywheel solution
depends to a considerable extent upon the longitudinal profile of the
pipeline as well as on the type of liquid being conveyed. Pipeline
systems which include flywheel protection and which are much
longer than this 1—2 km limit can be found. Often the flywheel forms
162
Flywheels
part of an overall package of measures which may include air valves or

feeder tanks positioned elsewhere in this system. The flywheel provides
some initial suppression of the transient while these extra measures
provide additional local surge alleviation.
All other things being equal, where treated water is being conveyed
and pressures must remain positive throughout to avoid contamination,
a larger size of flywheel will be necessary to provide adequate protection
than if a sewage or raw water pumping system is involved in which lower
minimum pressures and even vacuum pressures may be acceptable.
Sewage pumps usually operate at lower speeds than clean water
pumps. This also has a pronounced influence upon the size of flywheel
required. When selecting a pump sometimes the choice is between units
having different operating speeds and this can have a marked effect
upon necessary flywheel capacity.
11.5 Case study with different pump speed options

Consider the Scottish Water treated water rising main from Lawhead to
Silverburn. The length of main is 3.15 km. One phase of the scheme’s
development employed a DN 200 DI main to convey a flow rate of
around 34 litres/s. During design of the system, two options were
considered for the pumping plant at Lawhead where a solo duty
pump would operate with a second unit on standby. One option was
to use a WEIR PUMPS Ltd, UNIGLIDE pump type SDC 50/80 with
a two-pole motor and operating speed of 2900 rpm. A second possibility
was to use a four-pole motor running at 1470 rpm and coupled to a
WEIR PUMPS Ltd, DUOGLIDE pump type DRA 125/150. Both
options had a basic pump þ motor moment of inertia ¼ 0.54 kg.m2 .
Having slightly different performance curves, the DUOGLIDE has
the ability to produce a slightly higher flow rate than the UNIGLIDE.
Being a treated water transfer system it was important to maintain
positive pressures throughout the rising main even during transient
events such as a pumping failure. For the UNIGLIDE option, Fig. 11.2
shows that subsequent to trip of the duty pump, minimum piezometric
levels over most of the main fell below the pipeline level, indicating that
sub-atmospheric pressures had occurred. Using the DUOGLIDE pump
a fairly similar pattern of behaviour was predicted as shown in Fig. 11.3.
The maximum piezometric levels which followed flow reversal remained
comfortably within the allowable limits for the pipeline in both cases.
Addition of a flywheel was considered as a means of alleviating the
minimum pressures. For the UNIGLIDE pump option a flywheel having
163
Lawhead/Silverburn rising main. Max. and min. head after trip of Uniglide pump
350
300
250
Elevation (mAOD)
200
150
i.l. (mAOD)
100 hmax/0.0
hmin/0.0
50 hmax/2.7
hmin/2.7
0
0
150
300
450
600
750
900
1050
1200
1350
1500
1650
1800
1950
2100
2250
2400
2550
2700
2850
3000
3150
Chainage (m)
Fig. 11.2. Envelope curves for trip of UNIGLIDE pumpset
moment of inertia ¼ 2.7 kg.m2 was found adequate for the task. Figure
11.2 shows that both maximum and minimum piezometric levels were
improved with minimum head remaining above the pipeline elevation.
In the case of the DUOGLIDE option it was necessary to have a
flywheel with moment of inertia ¼ 13.5 kg.m2 before a similar degree
Lawhead/Silverburn rising main. Max. and min. head after trip of Duoglide pump
350
300
250
Elevation (mAOD)
200
150
i.l. (mAOD)
hmax/0.0
100
hmin/0.0
hmax/13.5
50 hmin/13.5
0
0
150
300
450
600
750
900
1050
1200
1350
1500
1650
1800
1950
2100
2250
2400
2550
2700
2850
3000
3150
Chainage (m)
Fig. 11.3. Envelope curves for trip of DUOGLIDE pumpset
164
Flywheels
of improvement was achieved. The resulting curves of maximum and

minimum piezomeric level are shown in Fig. 11.3.
The difference of flywheel inertia necessary to achieve the required
improvement is largely due to the difference in operating speed for
the two pumps. The amount of stored energy in the rotating masses
is roughly comparable.
Stored energy E is given by:
E ¼ 12 I!2 ð10:4Þ
In the case of the UNIGLIDE pumpset:
E ¼ 12 ð0:54 þ 2:7Þð2900 2 =60Þ2 ¼ 149 406 N:m
And for the DUOGLIDE pumpset:
E ¼ 12 ð0:54 þ 13:5Þð1470 2 =60Þ2 ¼ 166 352 N:m
Despite the substantial difference in operating speed the amount of
energy stored is broadly similar.
Velocity variations at the start of the main were as shown in Fig. 11.4.
The pump is tripped at time ¼ 1 s. It is noticeable that failure of the
DUOGLIDE pump (without flywheel) produces a more rapid decelera-
tion than for the UNIGLIDE (without flywheels) despite the basic
pump þ motor inertia being almost the same. The more gradual
Lawhead/Silverburn rising main. Velocities after trip

1.4
SDC/0.0
1.2 SDC/2.7
DRA/0.0
1.0 DRA/13.5
0.8
Velocity (m/s)
0.6
0.4
0.2
0
0.041
0.615
1.189
1.763
2.337
2.911
3.485
4.059
4.633
5.207
5.781
6.355
6.929
7.503
8.077
8.651
9.225
9.799
10.373
10.947
11.521
12.095
12.669
13.243
13.817
14.391
14.965
15.539
16.113
–0.2
Time (s)
Fig. 11.4. Velocities after pumping failure
165
Lawhead/Silverburn rising main. Heads after trip

350
300
250
Head (mAOD)
200
150
100 SDC/0.0
SDC/2.7
50 DRA/0.0
DRA/13.5
0
0.041
1.025
2.009
2.993
3.977
4.961
5.945
6.929
7.913
8.897
9.881
10.865
11.849
12.833
13.817
14.801
15.785
16.769
17.753
18.737
19.721
20.705
21.689
22.673
23.657
24.641
25.625
26.609
27.593
28.577
Time (s)
Fig. 11.5. Head variations after pumping failure
initial deceleration in the case of the UNIGLIDE is again a consequence

of the higher operating speed which gives the UNIGLIDE almost four
times the stored energy of the DUOGLIDE.
A similar effect is to be seen in the time variation of piezometric level
after trip at the start of the rising main (Fig. 11.5). For the DUOGLIDE,
head falls more steeply at first than for the UNIGLIDE. It will be noted
that the head downstream of the DUOGLIDE stabilises after this initial
fall. This is because the head downstream of the pump has now fallen
below the wet well level and the pump has started to turbine.
The variation of velocity below the DUOGLIDE actually increases by
a modest amount, for a time during this pump turbining event (Fig. 11.4).
Due to a more gradual decline in head after trip, the UNIGLIDE starts to
turbine much later and for a shorter time. Arrival of wave reflections from
Silverburn after 4.5 s starts to produce a further stage in the flow
deceleration, leading finally to flow reversal and check valve closure
after 7 s. Head starts to recover after around 7 s (Fig. 11.5). It is inter-
esting to note that if wet well level were lower, then the head down-
stream of the pump would have continued to fall, causing more
severe sub-atmospheric pressures along the pipeline.
When the flywheel is added, the flow deceleration is prolonged (Fig.
11.4) and the decline in piezometric level is also lengthened (Fig. 11.5).
Flow reversal occurs after 12 s for the UNIGLIDE and at almost 14 s for
166
Flywheels
the DUOGLIDE. The pumps do not turbine once the flywheels have
been added, as minimum head is maintained at higher levels than
those within the suction well.
11.6 Flywheels on a larger system

A second example concerns a larger system comprising parallel sewage
rising mains each over 6 km in length. The pipelines from Sharjah
Pumping Station (PS) No. 1 to the treatment works are DN 450 and
DN 700. These mains rise to a peak at chainage 4321 m and an air
valve is sited on each main at this location. Flow rate is an order of
magnitude greater than in the previous example. Speed of the WEIR
PUMPS Ltd, SWALLOWGLIDE type SR400DS was 980 rpm, with a
design head of 34 m. Each of the two duty pumps is rated at
390 litres/s. Suction well level was at 3.182 mASL. A minimum sub-
atmospheric pressure head of 5 mWG, during surge events, was
adopted. Pump moment of inertia was 5.371 kg.m2 and that of the
motor was 12.5 kg.m2 . The delivery main leaving the pumping station
was DN 700 with a bifurcation into the DN 450 and DN 700 mains
occurring a short distance from the pumping station.
After two duty pumps were tripped, Fig. 11.6 shows the piezometric
level just downstream of the pumping station, with head falling steeply
Sharjah PS No. 1 head d/s of PS after trip

35
0.0 kg.m2
30 255 kg.m2
25
20
Head (mASL)
15
10
0
0.103
2.781
5.459
8.137
10.815
13.493
16.171
18.849
21.527
24.205
26.883
29.561
32.239
34.917
37.595
40.273
42.951
45.629
48.307
50.985
53.663
56.341
59.019
61.697
64.375
67.053
69.731
72.409
75.087
77.765
80.443
–5
Time (s)
Fig. 11.6. Head variations following pumping failure
167
Sharjah PS No. 1 trip of pumps. Velocity d/s of PS

2.5
0.0 kg.m2
255 kg.m2
2.0
1.5
Velocity (m/s)
1.0
0.5
0
0.103
2.060
4.017
5.974
7.931
9.888
11.845
13.802
15.759
17.716
19.673
21.630
23.587
25.544
27.501
29.458
31.415
33.372
35.329
37.286
39.243
41.200
43.157
45.114
47.071
49.028
50.985
52.942
54.899
–0.5
Time (s)
Fig. 11.7. Velocity changes after pumping failure
until it is below the sewage level in the suction well. At this time the
pumps begin to turbine and piezometric level downstream of the
pumps is largely stabilised by continued flow from the suction well.
This continues until around 40 s have elapsed when flow finally
ceases and check valves close. Flow reversal along the mains produces
a modest head rise and a fairly regular oscillation.
Figure 11.7 shows the corresponding velocity variation in the DN
700 line leaving the pumping station. Flow declines in a stepped
manner, with each wave reflection from the treatment works producing
a further reduction in flow until check valve closure.
Notwithstanding the stabilising effect of the suction well and pumps’
turbining, Fig. 11.8 shows that the entire length of pipelines up until the
summit at chainage 4321 m is subject to severe sub-atmospheric pres-
sure. Operation of the air valves at this high point alleviates conditions
between this summit and the treatment works.
The minimum pressures predicted were unacceptable and a flywheel
having inertia ¼ 255 kg.m2 was added to each pump. Subsequent
predictions after failure of the pumps produced a more gradual fall in
piezometric level downstream of the pumps (Fig. 11.6), with
minimum head remaining above suction well level. Flow reversal
occurred around 56 s with pumps tripped at time ¼ 1 s. After check
valves shut, subsequent peak pressures were below the steady
168
Flywheels
Sharjah PS No. 1 max. and min. head following pumping failure DN 450 main
30
i.l. (mASL)
25 hmax/0.0
hmin/0.0
hmax/255
20
hmin/255
Elevation (mASL)
15
10
0
0
301.5
603.0
904.5
1206.0
1507.5
1809.0
2110.5
2411.9
2713.4
3014.9
3316.4
3617.9
3919.4
4220.9
4422.3
4725.2
5028.0
5330.8
5633.6
5936.4
6144.9
-5
–10
Chainage (m)
Fig. 11.8. Envelope curves after pumping failure for DN 450 main
pumping levels. Velocity variations in Fig. 11.7 show a smoother

deceleration with the flywheels added. Minimum piezometric levels
along the mains (Fig. 11.8) show an improvement with an almost
linear variation of head from the pumping station until the summit at
chainage 4321 m. Minimum pressure head was above the 5 mWG
allowable limit.
The air valves have acted as a secondary form of protection against
low pressures but at the price of admitting substantial quantities of
air to the mains. Figure 11.9 shows the anticipated quantities of air
drawn into each pipeline with and without flywheels. Including the
flywheels has the effect of reducing the volume of air by almost 2 m3
in each main. Since there is a relatively small difference in level
between the summit at chainage 4321 m and the level of sewage in
the pipeline leading up to the treatment works inlet, it was predicted
that much of the air admitted would remain in the pipelines. When
pumps are restarted, air will be purged and it is important to have
safe rates of expulsion if undesirable hydraulic transient effects are to
be avoided when the air valves shut.
This case has shown that the size of flywheel necessary can increase
considerably when flow rate, pipe length and diameter increase and
when pump speed is relatively low. For instance, if pump speed had
been 1470 rpm then the necessary moment of inertia of flywheel
would have been closer to 100 kg.m2 . Providing pump start measures
169
Sharjah PS No. 1 air volumes around ch. 4321 m after pumps are tripped
12
10
6
Volume (m3)
4 450/0.0
700/0.0
2 450/255
700/255
0
0.103
13.081
26.059
39.037
52.015
64.993
77.971
90.948
103.926
116.904
129.881
142.859
155.836
168.814
181.792
194.769
207.747
220.724
233.702
246.680
259.657
272.635
285.612
298.590
311.568
324.545
337.523
350.500
363.478
376.456
389.433
–2
Time (s)
Fig. 11.9. Air volumes in mains after pumping failure
will allow, this example has also illustrated how a flywheel can be
considered for larger schemes, acting in conjunction with other forms
of alleviation, to provide an overall package of protection measures.
11.7 Booster pump installations

In the case of a booster pump fitted with a flywheel, then following a
pumping failure hydraulic transient behaviour is benefited both
upstream and downstream of the pumping station. This is because
deceleration dV=dt is controlled at its source — that is, at the pump.
This changing rate of flow applies both upstream and downstream of
the pumping station. Along a suction main the effect of the flywheel
is to suppress the peak transient pressures which would otherwise
occur following a pumping failure.
11.8 Multi-pump installations

For multi-pump installations it is important to note that the amount of
protection varies in proportion with the number of operating pumps,
whereas the incremental increase in flow rate obtained by adding
each additional pump becomes smaller. Extreme hydraulic transient
pressures after a pumping failure may not necessarily follow failure of
170
Flywheels
the maximum number of operating pumps but may result from trip of a
lesser number of units.
11.9 Advantages of flywheels

Advantages of the flywheel solution are: relatively lower cost as far as
initial purchase is concerned; simplicity; control of the surge at its
source; quieter check valve closure; control of surging both upstream
and downstream in the case of a booster pump; and no restriction as
far as the type of liquid being conveyed.
Appendix Moment of inertia

A body in linear motion has a kinetic energy ¼ 12 mass velocity2 . Con-
sider a solid disk of material of density d having diameter D, thickness t
and rotating with angular velocity ! as shown in Fig. A11.1. A segment
of this disk is at a distance r from the centre and occupies an area
dr r d as shown in the figure. The kinetic energy dE of this elemen-
tary segment of material will be dE ¼ 12 mass velocity2 .
Alternatively, the kinetic energy can be expressed as:
dE ¼ 12 d dr r d tV 2
since, V ¼ !r kinetic energy can be written:
dE ¼ 12 d dr r d tð!rÞ2 ¼ 12 d t!2 r3 dr d
Integrating over the entire disk then, total kinetic energy of the rotating
body will be given by:
ÐÐ
E ¼ 12 d t!2 r3 dr d
With the limits of integration being, 0 r D=2 and 0 2, this
gives:
E ¼ 12 d t!2 14 ðD=2Þ4 4=2 ¼ 12 d t!2 D4 =32
or,
E ¼ 12 fd tD4 =32g!2 ¼ 12 I!2 ðA10:1Þ
where I is called the moment of inertia of the rotating mass. The
moment of inertia of a more complicated rotating body is obtained by
summing the individual moments for each part. In the case of an
electrically driven turbine pumpset this will comprise the rotor of the
171
Angle, dq
Solid disk of material

Velocity, V
r dr
D/2
Fig. A11.1. Definition sketch for deriving moment of inertia
electric motor, the shaft connecting the motor and pump and the
impeller of the pump, possibly including entrained liquid.
Power P is rate of working or,
P ¼ dE=dt ¼ 12 I d!2 =dt ¼ I! d!=dt ðA10:2aÞ
If speed ! is increasing — that is, the machine is running up to full
speed — then d!=dt is þve and rotational kinetic energy is increasing,
and when d!=dt is ve the machine is decelerating and stored energy is
decreasing. The reduced kinetic energy of rotation is partially trans-
ferred to the liquid being pumped and so pumping is continued for a
time after a pump is switched off or tripped.
Torque T is defined as:
T ¼ P=! ¼ I d!=dt ðA10:2bÞ
and is an alternative to using power as a variable.
172
12
Pressure vessels
Pressure vessels form one of the most versatile means of providing surge
protection at pumping stations. A wide range in vessel capacity can be
provided, allowing the vessel or vessels to be optimised to suit any
system. This chapter describes the principles of operation of a pressure
vessel and provides some examples of installations.
All forms of hydraulic transient protection endeavour to limit the
rates of change of flow, acceleration or deceleration within at least
part of a pipeline network. Such flow changes may be introduced to a
system by starting or stopping of pumps. One method of limiting the
rate of acceleration/deceleration of flow is to provide some means of
augmenting flow into the pipeline. This is the principle of operation
of a pressure vessel when it is installed on the pipeline (Fig. 12.1).
A mass of gas, usually air or nitrogen, is contained in the upper part
of the vessel which is usually cylindrical in shape and installed
either nearly horizontally or vertically. Occasionally a sloping vessel
may be included such as at Ross Priory on Scottish Water’s Loch
Lomond Scheme where four vessels inclined at 458 are used
(Fig. 12.2).
A pressure vessel may be used to improve minimum head conditions
and/or maximum transient pressures.
12.1 Modelling a pressure vessel

Behaviour of a pressure vessel is governed by the response of the gas
mass contained within the upper parts of the vessel. Commonly the
response of the gas is represented by the universal gas law:
pabs Vol=Tabs ¼ constant ð12:1Þ
173
Pressure vessel
Gas charge
M
Pipeline
Flow from upstream decreasing Downstream flow augmented by

outflow from vessel |dV/dt| reduced
Fig. 12.1. Simple pressure vessel installation

where pabs is absolute pressure, Vol is the volume of gas and Tabs is
absolute temperature.
12.1.1 Polytropic relationship

The process of expansion/compression of the gas mass is often replaced
by an approximate relationship of the form:
pabs Voln ¼ constant ð12:2aÞ
where n is an exponent usually taken to be in the range 1:0 n 1:4
although values outwith this range are possible. Behaviour represented
in this way is termed ‘polytropic’ and a number of technically important
processes can be approximately described by this equation, see for
Fig. 12.2. Inclined pressure vessels
174
Pressure vessels
example Spalding and Cole (1963). It is possible to find a value of n

which more or less fits a particular set of experimental results. It is
important to note that whereas the ratio of specific heats is a property
of a gas, the empirical exponent n is not and is dependent upon the pro-
cess being considered. n ¼ 1:0 represents an isothermal process with all
changes of pressure and volume taking place at a constant temperature.
This implies that all necessary heat transfers to and from the gas mass
take place concurrently with the changes of pressure and volume.
The value n ¼ 1:4 models an isentropic process without loss or gain
of heat by the gas mass from its surroundings. Neither n ¼ 1:0,
n ¼ 1:4 nor any other constant value of n is true and these values do
not necessarily represent the extremes of n. In the short term during
a transient event there will be insufficient time for heat transfer of
the extent required to maintain a constant temperature within the
gas mass. Over a longer time period, temperature within the vessel
will be restored to the prevailing ambient level as heat is gained or
lost to its surroundings. This is illustrated by the recording of water
level within the pressure vessels of the Loch Lomond scheme at Bal-
more (Fig. 12.3) after pumps are started.
During start-up, air is compressed and its temperature rises above
ambient conditions. When steady flow and pressure have been achieved
within about 5 min, the record shows a gradual rise in the vessels’ water
level as the overlying air mass cools. There is no doubt that the
exponent n is not constant during the overall process but use of
Auto-sequence start of 24 MGD pumps No. 4 and 5. Surge vessel level from static
1.2
Sluice valve level
1.0
Water level (metres from static level)
0.8
0.6
0.4
0.2
0
0
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
–0.2
Time (min)
Fig. 12.3. Water level in vessel following pump start
175
the polytropic relationship is relatively simple, hence its continued

popularity.
12.1.2 Rational heat transfer (RHT) equation

To address perceived shortcomings of the polytropic equation, Graze
et al. (1976) studied behaviour of an instrumented pressure vessel. Mea-
surements of pressure, air volume and temperature within the vessel
were gathered. These data showed large and important variations in
temperature including the development of freezing temperatures.
They concluded that the observed temperature changes played an
important role in performance of the vessel and that latent heat
could have a significant influence on behaviour. The rational heat
transfer (RHT) equation presented by Graze (1968) was used to
simulate performance of the air charge in the instrumented vessel
with improved predictions over the polytropic equation. The RHT
equation can be written:
dpabs =dt þ pabs =Vol dVol=dt þ ð 1Þ=Vol dQ=dt ð12:3Þ
where pabs is absolute pressure, Vol is gas volume and dQ=dt is the rate
of heat outflow considered to be entirely due to free convection, defined
as:
dQ=dt per unit surface area ¼ 1:4jtj1=3 t ð12:4Þ
with t being the differential temperature between the gas mass and
ambient conditions. Research into the influence of latent heat transfer
was considered worthwhile, with the possibility of further improvement
in model predictions resulting.
12.2 Role of a pressure vessel in surge suppression

Suppose a pressure vessel is installed downstream of a pumping station as
illustrated in Fig. 12.4. If the pump is in steady operation the gas mass in
the vessel will be compressed to a volume corresponding to the steady
pumping pressure at the connection point. If the pump is tripped, the pres-
sure downstream of the pump will start to fall and flow from the pump will
diminish. As the effect of falling pressure reaches the vessel then the pres-
sure within the mass of gas will also start to fall, with a corresponding
expansion of gas volume according to the assumed relationship:
p
Vol ¼ 1=n ðconstant=pabs Þ ð12:2bÞ
176
Pressure vessels
Steady pumping hydraulic gradient
Hydraulic gradient shortly after

pumping failure
Head developed by failing pump

shortly after trip
Pressure vessel
M
Small |dV/dt| downstream of
vessel connection
Check valve Rising main

Failing pump
Larger |dV/dt | upstream of vessel connection

Suction well or wet well
Fig. 12.4. Instantaneous hydraulic grade line after pump trip
This expansion of gas volume has to be accommodated inside the vessel

by liquid flowing out of the pressure vessel. Flow into the downstream
pipeline is thus augmented by this outflow from the vessel, allowing a
more gradual deceleration of flow at the start of a rising main while a
steeper rate of deceleration is developed upstream of the vessel connec-
tion (Fig. 12.4) than would occur if the vessel were absent. The gradual
expansion of the gas mass also produces a slower fall in pressure at the
start of the main.
12.3 Initial estimation of required pressure vessel volume
12.3.1 Graphical techniques

Determining a suitable size of pressure vessel might start by using one of
the graphical methods of establishing a preliminary volume. Such
methods have been described by a number of investigators including
Graze and Forrest (1974), Thorley and Enever (1979) and Tucker
177
and Young (1960). These charts allow an initial assessment of vessel

volume to be made for many of the more common cases.
12.3.2 Simple numerical method

In circumstances where charts may not be available and when a rapid
assessment of vessel parameters is required then approximate equations
can be used. Perhaps the simplest of these equations was described by
Manuel (1970).
Starting from the assumption that the flow downstream of the vessel
decelerates to zero over an interval t ¼ 20L=a and if mean outflow from
the vessel is say half of the initial steady flow Qo then this simple
formula for volume abstracted from the vessel can be written:
Volume ¼ 12 Qo t ¼ 12 Qo 20L=a ð12:5aÞ
If wavespeed a 1000 m/s for a fairly rigid pipeline then:
Volume 12 Qo 20L=1000 0:01Qo L
12.3.3 More detailed numerical assessment

If a more accurate initial assessment of volume is required, an appendix
to this chapter describes the derivation and use of a set of equations
which allow vessel volume to be estimated. These equations were
derived from use of the rigid column approach.
Equations (A12.5a) and (A12.5b) are for the frictionless case with
n ¼ 1 and n 6¼ respectively.
Volm ¼ Volp eo =fzð1 hm =zÞ þ hm lnðhm =zÞg ðA12:5bÞ
Volm ¼ Volp eo =fz½1 ðhm =zÞ1=n
þ hm =ð1 nÞ½ðhm =zÞð1=ð1 nÞÞ 1g ðA12:5cÞ

Definition of terms used can be found in the appendix. More compli-
cated equations including the effects of pipeline resistance are also
included in the appendix as equations (A12.9a) and (A12.9b). Also
included in the appendix are illustrations showing how these equations
can be used.
12.3.4 Subsequent investigations and criteria

Subsequently detailed computations would be conducted of pumping
failure to determine more accurate minimum downsurge conditions
178
Pressure vessels
and the maximum expanded gas volume. To this maximum gas volume
is added a factor of safety in order to determine the gross vessel capacity
necessary.
Fixing minimum transient pressures is of particular importance as far
as treated water pipelines are concerned. Ideally the minimum transient
pressure should remain positive throughout the pipeline system for a
treated water application to avoid risk of contamination, but pressure
may be allowed to fall to a lower level in a raw water or sewage
system. Unfortunately it is a sad fact that in many countries economic
constraints make it difficult to find money for surge suppression, let
alone meet the strict criteria of maintaining positive pressures in treated
water supply pipelines. In such countries, circumstances dictate that
minimum transient pressures become sub-atmospheric with depressing
regularity given the often unreliable state of power supplies.
Having estimated a gas charge for a pressure vessel, more detailed
computations can be carried out. These calculations may commence
from static conditions with pumps being operated in sequence, for
multi-pump installations, to allow time for the hydraulic transient
developed from a pump start to dissipate before the next duty pump
is operated.
12.4 Case study of a sewage pumping system

Consider a sewage pumping system comprising two duty pumps
delivering into parallel mains each around 5 km in length. The pipeline
exiting from the pumping station is DN 600 and downstream of the
vessel connection the pipeline bifurcates into DN 600 and DN 450
mains. When a pump is operated, initially it is much easier for flow to
enter the vessel than to proceed along one of the two parallel mains.
Figure 12.5 shows the velocity variations predicted during start of
two sewage pumps with a 30 s time delay between operation of the
first and second pumps. Outflow from the vessel is taken to be þve.
Velocity in the discharge branch downstream of pump No. 1 increases
to a maximum after about 1 s, with the velocity in the DN 600 pipe
upstream of the vessel connection following a similar trend. Flow into
the vessel, shown as riser velocity, also falls to a minimum, with
inflow being ve. The velocity in the pipelines at chainage 1.5 km
increases much more gradually. As the gas charge becomes pressurised
then inflow to the vessel diminishes and with it the flow from the pump.
The pump is now experiencing a greater downstream head than that
imposed by static conditions. As velocity increases in the downstream
179
Abu Hamour PS 44. Velocities during sequenced pump start

1.5
0.5
Velocity (m/s)
0
0.045
3.060
6.075
9.090
12.105
15.120
18.135
21.150
24.165
27.180
30.195
33.210
36.225
39.240
42.255
45.270
48.284
51.299
54.314
57.329
60.344
63.359
66.374
69.389
72.403
75.418
78.433
81.448
84.463
87.478
–0.5
pump 1
pump 2
–1 u/s of connection
riser
ch 1.5 km
–1.5
Time (s)
Fig. 12.5. Velocity at a vessel connection during sequenced pump start
pipelines the pump output increases once more and some flow leaves
the vessel with pressure in the gas mass falling and the gas volume
expanding. After 30 s the second pump is operated and a further
period of pressurisation of the gas charge in the vessel occurs but to a
lesser extent than when the first pump was operated. After about
Abu Hamour PS 44. Sequenced pump start. Max. and min. head along mains
45
40
35
Elevation (mQNHD)
30
25
20 i.l. (mQNHD, where QNHD = Qatar National Hydrographic Datum)

h (max.)
15 h (min.)
10
0
0
150
350
550
750
900
1100
1300
1500
1650
1850
2000
2200
2400
2550
2750
2900
3100
3300
3500
3700
3900
4100
4300
4500
4700
4904
Chainage (m)
Fig. 12.6. Envelope curves along a rising main after pump start
180
Pressure vessels
112 min essentially steady flow has been established in the pumping
system. It will be noted that the initial discharge from the first pump
exceeds its design steady output. This can have implications with
regard to the short-term power requirements for the pump. The overall
variation of piezometric level along the pumping mains is shown in
Fig. 12.6. Typically a vessel installation will produce a fairly uniform
variation of head along a main. As shown in the figure, inclusion of
the pressure vessel has produced a relatively smooth variation in max-
imum head.
12.5 Worst-case conditions

In attempting to establish a suitable capacity of pressure vessel it is
important to determine the ‘worst case’ as far as the demand on the
vessel is concerned. Typically, for a multi-pump installation, simulta-
neous trip of all duty pumps in steady operation would be examined.
Consider a treated water pumping scheme involving sequenced start
of three pumps at 10 s intervals with a pressure vessel installed down-
stream of the pumps. Figure 12.7 shows the predicted velocity varia-
tions in the DN 700 pipeline just upstream and downstream of the
vessel connection. Upstream of the vessel connection, flow from each
pump rises steeply and then levels off.
Downstream of the vessel connection a more gradual velocity increase
occurs. After the third pump has been operated, output from the pumping
Llandinum high-lift pumps. Sequenced start of three pumps
2.5
u/s of connection
d/s of connection
2
1.5
Velocity (m/s)
0.5
0
0.016
1.568
3.120
4.672
6.224
7.776
9.328
10.880
12.432
13.984
15.536
17.088
18.640
20.192
21.744
23.296
24.848
26.400
27.952
29.504
31.057
32.609
34.160
35.712
37.264
38.816
40.368
41.920
43.472
45.024
46.576
–0.5
Time (s)
Fig. 12.7. Flow development at a high-lift pumping station
181
Llandinum high-lift pumps

3.5
2.5
Air volume (m3)
1.5
1
n = 1.2
0.5 n = 1.0
n = 1.4
0
0.016
1.552
3.088
4.624
6.160
7.696
9.232
10.768
12.304
13.840
15.376
16.912
18.448
19.984
21.520
23.056
24.592
26.128
27.664
29.200
30.737
32.273
33.808
35.344
36.880
38.416
39.952
41.488
43.024
44.560
46.096
47.631
Time (s)
Fig. 12.8. Computed air volume as a function of polytropic coefficient
station upstream of the vessel varies only by a modest amount. In contrast,

the flow downstream of the vessel is augmented by flow leaving the pres-
sure vessel after a period when the gas charge has been over-pressurised
following start of the pumps. The maximum flow in the pipeline system
is for a time greater than the design maximum flow. Should a pumping
failure occur during this period of excess flow then the demand on the
vessel may be significantly greater than if pump trip were to take place
from a steady flow condition. Likewise, variations in gas volume within
the pressure vessel will be greater when pumps are tripped at this
maximum flow than if pump failure occurs under steady flow. It is not
suggested that simulations should always include failure from this tran-
sient maximum flow condition but that the possible implications be
borne in mind when choosing a margin of safety for vessel volume.
Uncertainty with regard to the behaviour of the gas charge in the
vessel has to be considered when sizing a vessel. Figure 12.8 illustrates
the changing gas volume predicted after all three duty pumps were
tripped from steady maximum discharge, using values of n ¼ 1:0, 1.2
and 1.4. The effect of changing the polytropic coefficient is evident
with the highest value of n producing the maximum vessel volume
requirement. A similar effect on both maximum and minimum head
is to be seen in Fig. 12.9. Selection of an appropriate vessel volume
has to include some consideration of any limitations in the means of
modelling gas behaviour.
182
Pressure vessels

300
250
200
Head (mAOD)
150
100
n = 1.2
50 n = 1.0
n = 1.4
0
0.016
1.552
3.088
4.624
6.160
7.696
9.232
10.768
12.304
13.840
15.376
16.912
18.448
19.984
21.520
23.056
24.592
26.128
27.664
29.200
30.737
32.273
33.808
35.344
36.880
38.416
39.952
41.488
43.024
44.560
46.096
47.631
Time (s)
Fig. 12.9. Predicted head as a function of polytropic coefficient
Whether starting from an estimated gas charge obtained from one of

the approximate methods, or simply from a first guess, the design pro-
cess usually will involve a series of analyses using alternative gas
charges. As gas charge (static air volume) is increased, the amplitude
of head change reduces with minimum piezometric level increasing
and maximum head decreasing along the pipeline. In many instances,
especially treated water systems, the optimum gas charge will be deter-
mined by the requirement to avoid too low pressures. The appropriate
gas charge is often dictated by conditions somewhere along the pipeline,
say at a local high point on the main. In the present case, minimum
pressure at a point around 800 m from the pumping station was the
critical factor in fixing vessel capacity. To maintain positive pressures
at this location requires a minimum static gas volume of 0.8 m3 and
to maintain þ0.2 bar g requires a static volume of 1.2 m3 . Vessel
volume can be quite sensitive to minimum pressure required.
12.6 Reversed flow and refilling a pressure vessel

Having successfully established a vessel gas charge to adequately con-
trol minimum pressures after pumping failure, flow will have come to
rest with a minimum hydraulic gradient along the pipeline. Under the
action of this usually adverse hydraulic gradient, flow will reverse and
accelerate back towards the pressure vessel which will start to refill.
183

300
250
200
Head (mAOD)
150
100
50 No throttle
DN 150 bp
0
0.016
1.600
3.184
4.768
6.352
7.936
9.520
11.104
12.688
14.272
15.856
17.440
19.024
20.608
22.192
23.776
25.360
26.944
28.528
30.113
31.697
33.281
34.864
36.448
38.032
39.616
41.200
42.784
44.368
45.952
47.535
Time (s)
Fig. 12.10. Head variations showing effects of throttling inflow to vessel
The adverse hydraulic gradient can have a magnitude as large or larger

than the steady pumping gradient and reversed velocities may be sub-
stantial. The strength of the reversed velocity developed will depend
upon the head available and the time over which the gradient acts.
As flow re-enters the pressure vessel the gas mass which had previously
expanded will now begin to be compressed and its internal pressure will
start to increase in accordance with the gas law. As gas pressure rises,
the adverse hydraulic gradient flattens and the rate of acceleration
towards the vessel then decreases. Ultimately pressure in the vessel will
reach a level at which no further acceleration occurs and the reversed
velocity starts to decrease. The gas mass continues to be compressed
until the reversed flow has reached zero. Pressure in the vessel is now at
its peak for this reversed flow event. Where system static head is
modest and the vessel volume is relatively large, the peak pressure after
reversed flow and vessel refilling may not reach steady pumping levels.
In other circumstances, particularly where static head is large and vessel
volume is more modest, the maximum pressure after reversed flow may
exceed steady pumping level by a considerable margin as illustrated in
Fig. 12.10.
In these schemes where the peak pressure is found to be acceptable
then the gas volume and vessel capacity established to control mini-
mum pressures can be also be considered adequate for the maximum
pressure after flow has refilled the vessel. On the other hand, if
184
Pressure vessels
Downstream piezometric level in service reservoir M
Piezometric gradient with

throttling at vessel inlet
Total head
available
Head drop
in throttle
Piezometric line
without throttling
Pumping main
Piezometric level in vessel

Reversed velocity
M
Pressure vessel
Pumping station
Fig. 12.11. Sketch showing throttle effect
maximum transient pressure is excessive, measures require to be taken

in order to reduce these pressures to within acceptable limits. Two
options present themselves.
(a) Increase the gas volume and vessel size to reduce pressure ampli-
tude generally.
(b) Alternatively, the reversed flow returning to the vessel and which
produces the peak pressure during compression of the gas mass,
can be ‘throttled’ to damp out or reduce the energy of the returning
flow as illustrated in Fig. 12.11. The act of throttling inflow to the
vessel after velocity reversal in the pipeline causes head in the main
at the vessel connection to increase in order to overcome the
resistance of the throttle. This reduces the magnitude of the
adverse hydraulic gradient which was responsible for acceleration
of the reversed flow (Fig. 12.11). Magnitude of reversed velocity
is thus limited, and with a smaller return flow the hydraulic level
in the vessel necessary to decelerate this flow is also reduced.
The peak pressure of the return upsurge is thus limited.
185
A throttling effect can be achieved in several ways but in all cases the
head drop H achieved by the throttle can be represented as:
X
H ¼ ðKL =A2 ÞQ2 =ð2gÞ ð12:6Þ
where KL is the loss coefficient of a component, A is the cross-sectional

area of the component and Q is flow rate through the component.
Options for the throttle are as follows.
(a) When a modest amount of throtting is required a differential orifice
(Fig. 12.12(a)) may be an attractive solution as no moving parts are
involved. The amount of resistance offered by the differential
throttle is fixed by its geometry.
(b) A set of tapers (Fig. 12.12(b)), could be used to produce a larger
head loss than with the single differential orifice. This has been
used successfully in a number of applications as described by
Popescu (1985).
(c) A non-return valve could be installed in the vessel connection and
an orifice formed in the valve door (Fig. 12.12(c)). The valve would
be orientated to open during outflow from the vessel. After flow
reversal the valve shuts and reversed flow is forced through the
orifice with attendant head loss. The orifice can be sized to produce
the necessary amount of throtting provided the jet passing through
the orifice does not have excessive velocity. Orifice size fixes the
amount of resistance. Head loss through this arrangement can be
appreciably greater than for the differential orifice.
(d) A more versatile throttle can be arranged using a bypass around a
non-return valve, again placed in the vessel connection so as to
open on outflow (Fig. 12.12(d)). The bypass diameter will usually
be appreciably smaller than for the vessel connection. Various
elements such as a gate valve can be placed in the bypass and set
to a part-open position. Other fixed elements such as orifice
plates can also be installed in the bypass to add extra throttling
where the necessary head drop is not achievable without a very
small opening in the gate valve. The orifices can be used to
remove much of the excess head, leaving the valve to ‘tune’ the
overall throttling effect. Large amounts of head loss can be achieved
using this arrangement.
(e) Linser (2004) described an innovative throttle devised in 1932 by
Thoma (see Linser, 2004). Almost unrestricted flow is allowed in
one direction through a cone shaped connection (Fig. 12.12(e)),
while water is set in centrifugal motion using a spiral casing type
186
Pressure vessels
Outflow from vessel
Outflow from vessel (1.67 – 2.5) × Do
Do
(3.34 – 5.0) × Do
Do
2 × Do
Orifice may be shaped onto the
vessel base or held between pipe flanges (1.67 – 2.5) × Do
(a) (b)
Bypass containing valve and

possible orifice plate(s)
Check valve open

during outflow
Orifice in valve door
Outflow from vessel Check valve closed Outflow from vessel Check valve
during inflow
(c) (d)
Helicoidal inflow
Spiral casing
Pressure vessel
Isolating valve Relatively unrestricted outflow
Rising main
(e)
Fig. 12.12. Types of throttle
arrangement at inlet to the connection when flow is in the opposite

direction. The device has the advantage of no moving parts. Used
in the Kauner Valley hydroelectric plant to throttle surge tank
flows, resistance in centrifugal motion was 50 times that in the
unrestricted flow direction.
187

300
250
200
Elevation (mAOD)
150
100 i.l. (mAOD)

hmax-no
hmin-no
50 hmax-150
hmin-150
0
0
36.88
73.76
110.64
147.52
184.40
221.28
258.16
295.04
331.92
368.80
405.68
442.56
479.44
516.32
553.20
590.08
626.96
663.84
700.72
737.60
774.48
811.36
848.24
885.12
922.00
Chainage (m)
Fig. 12.13. Envelope curves showing throttle effect after a pumping failure
Figure 12.10 demonstrated the influence upon piezometric level at

the vessel connection, of throttling inflow to the pressure vessel using
a check valve and bypass arrangement. The check valve and vessel con-
nection are DN 300 and the bypass is DN 150. Maximum head has
been reduced by about 30 m. The influence on maximum piezometric
levels along the pipeline is shown in Fig. 12.13. Minimum head is
unaffected by the throttle which does not operate during outflow. Mini-
mum air volume within the pressure vessel is influenced by the throttle.
Since maximum pressure is reduced, the air mass is not recompressed by
the returning flow to the same extent. Maximum air volume is not
affected.
Pressure vessels can be made in a wide range of sizes. Smaller
capacities are available off the shelf while larger volumes ranging up
to about 250 m3 can be custom-made. Probable 250 m3 is about the
practical upper limit as far as transportation to most sites is concerned.
Where a still larger capacity is necessary this can be made up of several
vessels.
Some pumping stations may have total pressure vessel volumes of
around 1000 m3 installed, for example at Scottish Water’s Ross Priory
and Balmore pumping stations.
Pumping failure in the presence of a pressure vessel located down-
stream of a high-lift pumping station is usually characterised by rapid
188
Pressure vessels
flow reversal and closure of check valves on pump delivery branches.

Once the check valves have closed, conditions on the upstream side
of the valves play no further part in the transient event within the
downstream pumping system. This was the situation for the Llandinum
high-lift pumping system.
12.7 Low-lift systems

Many systems operate with much lower static and pumping heads with
conditions upstream of pump check valves playing a more important
role over a longer time period.
Consider the case of the transmission system from Rifa’a Blending
Station in Bahrain to Hamad Town. The water level in distilled
water storage facilities at Rifa’a may vary from a minimum or bottom
water level (BWL) of 49.15 mBNSD (Bahrain National Survey
Datum) to a maximum or top water level (TWL) at 57.35 mBNSD.
At the Hamad Town tanks, the water level can range from a
BWL ¼ 42.5 mBNSD to a TWL ¼ 50.55 mBNSD. Since discharge
into the Hamad Town storage tanks is at TWL, the range in level
within the tanks does not affect hydraulic conditions along the twin
DN 600 DI mains. These mains are around 5.4 km in length and
have an undulating profile with a low point at Buri junction, chainage
2.6 km, from the pumping station.
Each pump has a rated duty of 280 litres/s at a head of 29 m. Pump
speed is 1490 rpm and moment of inertia of each pumpset is 1.7 kg.m2 .
A maximum of two duty pumps may operate at any time and pumping
failure can occur at any suction well level, thus requiring investigation
of the full range in head conditions. A relatively small pressure vessel
was installed on the delivery side of the pumping station. Figure
12.14 shows the predicted variation of piezometric level, mBNSD, at
Rifa’a for a pumping station blackout with suction tank level at BWL
and at TWL. An initial steep fall in head occurs downstream of the
pumps. This fall is arrested when the discharge level becomes less
than the suction tank level. Flow then continues through the pumping
station and it is this flow which is responsible for maintaining hydraulic
levels. The pressure vessel’s primary function is to attenuate the initial
part of the head drop and it plays only a small part in subsequent events.
When head is at TWL the system is under negative static head and
some flow can continue under gravity. The pumps were essentially aug-
menting the gravity flow. Conditions at TWL become steady after about
1 min. At BWL a small positive static head exists which gradually brings
189
Rifa'a/Hamad Town. Head variations at PS

70
60
50
Head (mBNSD)
40
30
20
Head (TWL)
10
Head (BWL)
0
0.090
8.910
17.730
26.550
35.370
44.190
53.010
61.830
70.650
79.469
88.289
97.109
105.928
114.748
123.568
132.387
141.207
150.027
158.846
167.666
176.486
185.305
194.125
202.944
211.764
220.584
229.403
238.223
247.043
255.862
264.682
Time (s)
Fig. 12.14. Head variations in a low-lift system after pump trip

flow to rest and produces a modest reversed velocity. This flow reversal
produces the small rise in piezometric level, peaking after about 215 s as
shown in Fig. 12.14.
Continuing flow through the pumping station plays an important role
in minimising the required vessel capacity. Figure 12.15 shows the
Rifa'a/Hamad Town. Velocity variations
1.6
d/s (BWL)
1.4
u/s (BWL)
d/s (TWL)
1.2
u/s (TWL)
1.0
Velocity (m/s)
0.8
0.6
0.4
0.2
0
0.090
8.910
17.730
26.550
35.370
44.190
53.010
61.830
70.650
79.469
88.289
97.109
105.928
114.748
123.568
132.387
141.207
150.027
158.846
167.666
176.486
185.305
194.125
202.944
211.764
220.584
229.403
238.223
247.043
255.862
264.682
–0.2
Time (s)
Fig. 12.15. Velocity variations in a low-lift system after pump trip
190
Pressure vessels
Rifa'a/Hamad Town. Air volume variations

7
5
Volume (m3)
1 Vol. (BWL)
Vol. (TWL)
0
0.090
8.730
17.370
26.010
34.650
43.290
51.930
60.570
69.210
77.850
86.489
95.129
103.768
112.408
121.048
129.687
138.327
146.967
155.606
164.246
172.886
181.525
190.165
198.805
207.444
216.084
224.724
233.363
242.003
250.643
259.282
267.922
Time (s)
Fig. 12.16. Air volumes in low-lift vessels after pump trip
changing velocity upstream and downstream of the pressure vessel

connection after the pumping failure at time ¼ 1 s. After pumps are
tripped, velocity falls sharply and check valves close. Downstream of
the vessel connection, velocity changes much more slowly as the
vessel starts to supply water to the pipeline system. When the piezo-
metric level downstream of the check valves becomes less than the
suction tank level, the check valves reopen and flow through the pump-
ing station is re-established with flows upstream and downstream of the
vessel’s connection becoming essentially the same after a relatively
short time. This indicates that the vessel is no longer supplying water
to the system. With a high suction tank level and negative static
head, a gravity flow is established. With lower tank levels and positive
static head, velocity gradually decelerates, with flow reversing after 212
min at BWL. The check valves now close. A small reversed flow
occurs and a modest repressurisation of the vessel air charge occurs.
Flow then becomes positive once more and as head falls the check
valves may again reopen.
Air volume variations in the pressure vessel are shown in Fig. 12.16.
After the initial pumping failure, volume expands to a maximum. When
check valves reopen and flow is re-established through the pumping
station, air volume decreases by a modest amount. Under negative
system static head conditions — that is, high suction tank levels
(TWL) — air volume is stabilised by the gravity flow condition. For
191
Rifa'a/Hamad Town. Trip of two pumps

70
60
50
Elevation (mBNSD)
40
30
i.l. (mBNSD)
20 max. (BWL)
min. (BWL)
10 max. (TWL)
min. (TWL)
0
0
204
407
611
814
1018
1221
1425
1629
1832
2036
2239
2443
2646
2850
3053
3257
3461
3664
3868
4071
4275
4478
4682
4886
5089
5293
Chainage (m)
Fig. 12.17. Envelope curves in a low-lift system after pump trip
positive static with the suction tank at BWL, a modest recompression

under reversed flow occurs.
The extremes of piezometric level along the 5.4 km pipelines are
shown in Fig. 12.17. Maximum hydraulic gradients are those of
steady pumping flow since there is either no return flow upsurge or
only a modest upsurge. With TWL in the Rifa’a tanks, both flow rate
and also hydraulic levels are at a maximum. The minimum head over
most of the mains is largely dictated by the prevailing suction tank
level at Rifa’a. Only towards the Hamad Town end of the system
does the influence of the higher flow rate at TWL produce a more
significant fall in head.
It is often the case within low static head systems, after pumping fail-
ure, check valves remain open for prolonged periods, or reopen, possibly
on several occasions. In the absence of a significant positive static head
to produce a strong reversed velocity, large head rise at the vessel is
often avoided with peak head following flow reversal not reaching
steady pumping levels over most of a pipeline. The size of vessel may
be reduced if appreciable quantities of water continue to pass through
the pumping station after a power failure. Note that if trip occurs
with a low suction tank level, sufficient water must remain in the
tank to provide the necessary continued supply otherwise air may
start to be drawn into pump suction branches.
192
Pressure vessels
For the example illustrated, the pipeline profile may be described as

favourable from the point of view of avoiding sub-atmospheric pressure
conditions. In other instances such as where a higher summit occurred
part-way along the main, the vessel volume would require to be much
larger.
12.8 Vessels at a booster pumping station

Hydraulic transient conditions at a pumping station can also be influ-
enced by the presence of a suction main. Some aspects of possible
behaviour are illustrated in the following example which involves two
pumping stations. Figure 12.18 shows the longitudinal profile and the
steady flow hydraulic gradient of a pipeline system. A lengthy suction
main extends from Yarker Bank reservoir at chainage 0.0 km for more
than 20 km, where the first pumping station is sited at Bainbridge.
Downstream of Bainbridge pumping station delivery head is in excess
of 350 mAOD. At around 27 km is Stonehouse pumping station
where head is lifted once more, allowing water to reach Fossdale service
reservoir (SR). The suction main bifurcates at Bainbridge pumping
station supplying pumps serving the two mains and a set of duty
pumps on each branch delivers treated water into rising mains serving
Fossdale SR and Aysgarth SR. The Bainbridge to Aysgarth main has an
overall length of 9.1 km. Pump delivery head of around 300 mAOD is
Wensleydale Project. Yarker Bank to Fossdale

400
350
300
Elevation (mAOD)
250
200
150
100
i.l. (mAOD)
50
Hydraulic gradient
0
0
977
1954
2932
3909
4886
5863
6840
7818
8795
9772
10 749
11 726
12 703
13 681
14 658
15 635
16 612
17 589
18 567
19 544
20 195
20 883
21 968
23 053
24 139
25 224
26 309
27 026
28 090
Chainage (m)
Fig. 12.18. Steady flow hydraulic gradient in a booster pump system
193
Fig. 12.19. Insulated pressure vessels
considerably lower than that of the pumps serving the main to Fossdale
SR. This difference in pump discharge head for the two mains has a sig-
nificant role to play in determining hydraulic transient behaviour.
Pressure vessels are located downstream of both the Fossdale and
Aysgarth pumps at Bainbridge and also upstream and downstream of
Stonehouse pumping station. Figure 12.19 shows these vessels which
are fitted with insulation jackets to avoid freezing in winter.
12.8.1 The upstream pumping station

Considering a station blackout at Bainbridge affecting both sets of
pumps, Fig. 12.20 illustrates the predicted variations in transient
head just upstream of the pumping station and downstream of both
pumps. After pumps are tripped at 10 s, Fig. 12.20 shows head in the
suction main rising steeply. Piezometric level downstream of the
Fossdale pumpset falls gradually under the influence of the pressure
vessel while head downstream of the Aysgarth pump also falls smoothly
194
Pressure vessels
Wensleydale Project. Heads at Bainbridge

400
350
300
250
Head (mAOD)
200
150
100 Suction
Aysgarth
50 Fossdale
0
0.270
9.180
18.090
27.000
35.910
44.820
53.730
62.640
71.550
80.460
89.370
98.280
107.190
116.099
125.009
133.919
142.830
151.740
160.650
169.560
178.470
187.380
196.290
205.201
214.111
223.021
231.931
240.841
249.751
258.661
267.571
Time (s)
Fig. 12.20. Head variations after trip at booster station
but at a greater rate due to the smaller vessel on this line. When head
on the suction side of the station exceeds head downstream of the
Aysgarth pump, that pump starts to turbine with flow passing through
the pump from upstream. Suction head and downstream head follow
one another closely. Head downstream of the Fossdale pump remains
unaffected by the suction level.
A compression wave travels upstream in the suction main to Yarker
Bank and is reflected back to Bainbridge arriving after about 45 s when a
steep fall in suction head occurs. The fall in head causes flow into the
Aysgarth line to cease and its pump check valve to close. Head down-
stream of the Aysgarth pump then falls smoothly as its vessel air volume
expands. Subsequent further wave reflections in the long suction main,
coupled with falling head downstream of the Aysgarth check valve,
allows this valve to reopen with a further period of flow through the
valve from upstream occurring. This sequence is repeated on one
further occasion.
Figure 12.21 shows the predicted variation of air volume within the
Aysgarth vessel. After trip at time ¼ 10 s, air volume starts to expand
but after a short interval volume is stabilised by the redevelopment of
flow through the pump. Wave reflection in the suction main causes
upstream head to fall and the check valve to close after about 45 s.
The Aysgarth main now becomes entirely reliant upon the vessel for
a continued supply of water. Air volume in the pressure vessel then
195
Wensleydale Project. Air volume/Aysgarth vessel

0.45
0.40
0.35
0.30
Volume (m3)
0.25
0.20
0.15
0.10
0.05 Volume
0
0.270
9.180
18.090
27.000
35.910
44.820
53.730
62.640
71.550
80.460
89.370
98.280
107.190
116.099
125.009
133.919
142.830
151.740
160.650
169.560
178.470
187.380
196.290
205.201
214.111
223.021
231.931
240.841
249.751
258.661
267.571
Time (s)
Fig. 12.21. Air volume variations after pump trip at booster station
expands to its maximum. Subsequent reopening of the check valve at
around 80 s produces a reduction in vessel volume as head rises and
flow passes through the Aysgarth pump and check valve. Thereafter
only modest variations in vessel volume were predicted to occur.
The development of pump turbining action has been beneficial in
two respects.
(a) When head upstream of the pump exceeds the downstream head
and the pump commences to turbine, this acts as a form of ‘relief
valve’ allowing water to escape from the pressurised suction main
and preventing further significant head rise on the upstream side
of the pumping station.
(b) The continued flow through the pumps while turbining means that
the downstream pipeline is not entirely dependent upon the pres-
sure vessel for a supply of water to achieve hydraulic transient alle-
viation but some of the water is being provided by flow continuing
to pass through the pump during the intervals of turbining.The
capacity of pressure vessel can be made smaller than if no turbining
action occurred.
12.8.2 The downstream pumping station

Considering the second pumping station at Stonehouse, the suction
main at this station is also the discharge main of the upstream pumping
196
Pressure vessels
Wensleydale Project. Head at Stonehouse

400
350
300
Head (mAOD)
250
200
150
100
u/s head
50
d/s head
0
59.890
64.395
68.900
73.405
77.910
82.415
86.920
91.425
95.930
100.440
104.940
109.450
113.950
118.460
122.960
127.470
131.970
136.480
140.980
145.490
149.990
154.500
159.000
163.510
168.010
172.520
177.020
181.530
186.030
190.540
Time (s)
Fig. 12.22. Head variations at booster station after all pumps are tripped
station. Events at one station will have an important influence upon

conditions at the neighbouring station. For instance, if a pump at one
of the stations is tripped this may require a pump at the second station
to be tripped after a time. Alternative conditions at the upstream Bain-
bridge station will be used to illustrate the changing effect on behaviour
at the downstream Stonehouse station. Relatively small pressure vessels
of identical gross volume but containing differing air charges and oper-
ating at different pressures were installed upstream and downstream of
the duty pump. These vessels were of the bladder type as described in
Chapter 13.
In the first of these illustrations the upstream pumps at Bainbridge
pumping station were tripped and the Stonehouse pump continued in
operation for a further 74 s before being tripped. During the intervening
period, suction head at Stonehouse gradually declined as illustrated in
Fig. 12.22. When the Stonehouse pump was switched off, suction head
started to rise smoothly, influenced by the upstream pressure vessel. On
the delivery side of the station, head started to fall smoothly controlled
by the downstream pressure vessel. The increasing suction head reaches
the downstream head which is falling at time ¼ 95 s. At this time the
check valve reopens and flow occurs from the suction side to the dis-
charge side of the pump. The upstream and downstream heads are
quite similar until around 100 s when the falling upstream head
causes the check valve to shut and flow through the pump to cease.
197
Wensleydale Project. Head at Stonehouse

450
400
350
300
Head (mAOD)
250
200
150
100
u/s head
50 d/s head
0
0.265
2.385
4.505
6.625
8.745
10.865
12.985
15.105
17.225
19.345
21.465
23.585
25.705
27.825
29.945
32.065
34.185
36.305
38.425
40.545
42.665
44.785
46.905
49.025
51.145
53.265
55.385
57.505
59.625
61.745
63.865
65.985
Time (s)
Fig. 12.23. Head variations at downstream station while upstream pumps are
running
Thereafter heads upstream and downstream vary independently with

the check valve remaining closed. The different periods of oscillation
upstream and downstream of Stonehouse pumping station are indica-
tive of the relative lengths of suction and delivery mains.
As far as velocity variations are concerned, after pump trip at 75 s,
velocity through the pump and check valve decelerates rapidly until
check valve closure. The check valve is assumed to have a good closure
performance. In the suction main upstream of the vessel, velocity
decelerates gradually, controlled by the upstream pressure vessel. In
the delivery main below the downstream vessel, velocity also decreases
relatively gradually under the influence of this vessel until 81 s when
flow reverses and some refilling of the vessel occurs. When upstream
head exceeds downstream head after about 95 s the check valve
reopens and there is a modest flow through the pumps until around
100 s when the check valve shuts once more.
In the second scenario the Stonehouse pump is tripped after 1 s while
the upstream Bainbridge pump continues to operate so that suction
head at Stonehouse is at a maximum during the event. Figure 12.23
shows the resultant variations in head upstream and downstream of
the Stonehouse pumping station. Head on the suction side of the
pump rises smoothly influenced by the upstream vessel while the
198
Pressure vessels
delivery head declines gradually controlled by the downstream vessel.

As upstream and downstream heads approach one another after
about 6 s, the check valve reopens and flow is re-established through
the pump. The suction and delivery heads are closely linked thereafter
with a damped oscillation occurring.
As far as velocities are concerned, following pump trip velocity
through the pump decelerates rapidly with check valve closure follow-
ing. Velocities in the suction and delivery mains again decelerate more
gradually, controlled by their respective pressure vessels. Flow reverses
in the delivery main after about 712 s and becomes positive once more at
around 14 s. The suction main velocity remains positive throughout. At
about 6 s the check valve reopens and flow passes through the pump.
The check valve remains open thereafter. As time passes, the velocities
gradually approach a new smaller steady flow condition under the
action of the Bainbridge pump alone.
This case demonstrates that a wide range of behaviours can occur at a
booster station, ranging from little or no turbining action at a failed
pump to the development of a new steady flow through the pump
which has been switched off. The suction and delivery main vessels
are instrumental in controlling rates of change of velocity and head in
the mains but at the price of producing a rapid deceleration of flow
through the failing pump. This high rate of deceleration carries with it
the risk of check valve slam if a suitable closure performance is not
achieved by the selected valve. The response of check valves is discussed
in detail in Chapter 20.
12.9 Summary of response with a pressure vessel included

A booster pumping station will have appreciable lengths of pipeline
both upstream and downstream of the pumps. There may be as great
a need for pressure transient protection of the suction main as there
is for a rising main.
Where a pressure vessel is placed on the delivery side of a pumping
station it can be shown that the rate of flow deceleration through the
pumps and their associated check valves is increased. If a suction
main is present then the increased rate of deceleration will impact
adversely on pressure transients in this main, tending to make surging
more severe. If a pressure vessel is then installed on the suction side
of the pumps in order to suppress these transient effects it will act to
receive water following a pumping failure. The pressure in the upstream
vessel will rise while that in the downstream vessel falls. The ability of
199
water to flow into the suction vessel rather than continue to the pumps
has the effect of further aggravating rates of flow deceleration through a
failing pump thus imposing more severe closure conditions on the pump
discharge check valves.
As the transient progresses it is not uncommon for the head upstream
of the pumps to exceed the downstream head. Then the check valves
will reopen, allowing flow to re-establish and also largely stabilising air
volumes within the vessels. Following flow reversal upstream and down-
stream of the pumping station, the upstream head will start to fall while
downstream head rises. Check valves will then reclose.
In the case of high-lift pumps, the check valves may not reopen if the
head upstream of the pumps does not rise to the minimum level of fall-
ing head in the downstream vessel. In these circumstances the more
rapid closure of the check valves will mean that more water will
enter the upstream vessel and more water will be withdrawn from the
downstream vessel. There will be a tendency for the necessary vessel
volumes to increase above the volumes required where a vessel was
installed only on one side of a pumping station.
As with a pressure vessel located on the downstream side of a pump-
ing station, throttling of flow can also be applied to the upstream vessel.
This throttling action may be applied either during inflow to the vessel
or to the outflow depending upon a need to suppress the upsurge on the
suction side (throtting of inflow) or the downsurge (throtting of out-
flow).
A pressure vessel may also be used as part of an overall protection
package. It can be used to prolong the time taken for piezometric
level to fall, at the start of a main, to a sufficient extent that no
unacceptably low pressures are developed between downstream points
on the main where other forms of protection are present.
Appendix Equations for estimating air vessel parameters

When a detailed hydraulic transient analysis of a pipeline system has
been commissioned and the time and costs associated with constructing
an accurate model have been covered then it is not so important to
have an initial estimate of pressure vessel characteristics. In other
instances, for instance where a tender is being prepared and a sum is
to be included for pressure transient protection, it is useful to be able
to price the pressure vessel installation as accurately as possible.
Should the contract not be awarded, any expense may not be recover-
able and so analysis time should be as brief as possible. The use of design
200
Pressure vessels
charts can be very helpful in this respect. In other circumstances, for

example during a meeting to discuss a range of aspects of a project,
questions may be raised unexpectedly about transient protection and
its likely costs. If charts for vessel selection are not to hand then it
may be appropriate to resort to an analytical approach as described
below.
There are essentially three initial questions that may need to be
answered for a proposed pressure vessel installation:
(a) What is the necessary capacity of pressure vessel and corresponding
gas volumes?
(b) What will be the peak pressure during the upsurge after flow
reversal?
(c) If the peak pressure from (b) is excessive, what degree of throttling
at the vessel inlet is necessary to reduce the peak pressure to an
acceptable extent?
The primary function of the vessel installation is to reduce rates of
deceleration and acceleration of the water column in the pipeline.
Assuming a relatively ‘slow’ transient event, then as a first approxima-
tion incompressible flow theory can be used allowing simpler physics
than the ‘elastic’ analysis. In common with the use of design charts to
estimate vessel characteristics, the water is assumed incompressible
and the conduit walls rigid.
The following information is necessary to conduct an assessment:
(a) pipeline profile or ground-level elevation along the route
(b) length and diameter of main
(c) design flow rate(s)
(d) maximum and minimum allowable pressures
(e) resistance to flow.
Vessel volume will often be determined by the need to control mini-
mum pressures during the initial downsurge, with peak pressures during
the return upsurge possibly requiring incorporation of a throttle device
to control inflow to the vessel.
A12.1 Equation of motion

Consider the pipeline system fitted with a pressure vessel as shown in
Fig. A12.1. It is assumed that when pumps are tripped they cease to
deliver flow instantaneously. A gas law of the form pabs Voln ¼
constant, is applicable, where:
201
Maximum allowable hydraulic gradient
habs
M
Transient hydraulic gradient

habs
z h
Minimum hydraulic gradient:
no vacuum allowed
vacuum allowed
Gas volume = Vol
Pipeline length = L and diameter = D
Pumping station
Fig. A12.1. Definition sketch for simplified vessel analysis
habs ¼ atmospheric pressure head metres water

h ¼ the absolute pressure head in the vessel metres water
z ¼ d/s reservoir level pressure vessel level þ habs ðmÞ
¼ density of water ðkg=m3 Þ
g ¼ acceleration due to gravity ðm=s2 Þ
L ¼ length of pipeline (m)
D ¼ pipe diameter (m)
A ¼ cross-section area of pipe ¼ D2 =4
V ¼ velocity of flow ðm=sÞ
Vol ¼ gas volume in vessel ðm3 Þ
X
F ¼ overall resistance factor ¼ fL=D þ KL
Suffixes:
o ¼ initial steady flow rate
m ¼ minimum head condition
p ¼ peak head condition
202
Pressure vessels
With reference to Fig. A12.1, the minimum acceptable head hm at

the pressure vessel during the downsurge after pumping failure is
obtained by drawing a straight line from the downstream reservoir to
the pressure vessel location, such that at no point along the line is
head less than the prescribed minimum. The atmospheric pressure
head hatm is added to the gauge pressure head to give the minimum
allowable absolute pressure head. If treated water is being carried
then the minimum acceptable head should remain positive, say at
least þ1 m above ground level. If sub-atmospheric pressures are accep-
table then this minimum head line will be lower as illustrated.
Mass of water in the pipeline ¼ AL
Acceleration of water column ¼ dV=dt
Force acting ¼ gAðh zÞ
Force due to flow resistance ¼ FV2 =ð2gÞgA
Since force ¼ mass and acceleration,
gAðh zÞ ¼ AL dV=dt FV 2 =ð2gÞgA
simplifying
L=g dV=dt FV2 =ð2gÞ ¼ h z ð2:1Þ
A12.2 Solution ignoring resistance to flow

Considering a frictionless pipeline in the first instance, then,
L=g dV=dt þ z h ¼ 0
or
L=g dV=dVol dVol=dt þ z h ¼ 0
but
dVol=dt ¼ AV
therefore,
L=gAV dV=dVol þ z h ¼ 0
or
LA dðV 2 =2gÞ=dVol þ z h ¼ 0 ðA12:1Þ
203
where LA ¼ Vp is the pipe volume and V 2 =ð2gÞ ¼ e is the kinetic

energy of flow. The absolute pressure head h can be represented by:
hVoln ¼ constant, or h ¼ constant Volð nÞ
Substituting then,
Vp de=dVol þ z constant Volð nÞ ¼ 0
or
Vp de þ z dVol constant Volð nÞ dVol ¼ 0 ðA12:2Þ
Integrating between the limits em and e and Volm and Vol then,
ð ð ð
Vp de þ z dVol constant Volð nÞ dVol ¼ 0
or
Vp ðe em Þ þ zðVol Volm Þ constant lnðVol=Volm Þ ¼ 0
for n ¼ 1 ðA12:3aÞ
and
Vp ðe em Þ þ zðVol Volm Þ
constant=ð1 nÞðVolð1nÞ Volð1nÞ

m Þ¼0 for n 6¼ 1
ðA12:3bÞ
Noting that em ¼ 0 and rearranging these equations, kinetic energy e
can be expressed as a function of changing volume Vol within the
vessel, thus:
e ¼ fzðVolm VolÞ þ constant lnðVol=Volm Þg=Vp for n ¼ 1
ðA12:4aÞ
and
e ¼ fzðVolm VolÞ þ constant=ð1 nÞ½Volð1nÞ Volð1nÞ
m g=Vp
for n 6¼ 1 ðA12:4bÞ
If the values e and Vol are taken as the initial steady flow conditions
represented by suffix o, equations (A12.4a) and (A12.4b) can be written:
Vp eo þ Volm zðVolo =Volm 1Þ Volm hm lnðVolo =Volm Þ ¼ 0
for n ¼ 1
204
Pressure vessels
and
Vp eo þ Volm zðVolo =Volm 1Þ
Volm hm =ð1 nÞðVolo ho =Volm hm 1Þ ¼ 0 for n 6¼ 1
The equations can also be arranged to allow solution for Volm , noting
that for the frictionless case ho ¼ z, thus,
Volm ¼ Vp eo =fzð1 hm =zÞ þ hm lnðhm =zÞg for n ¼ 1 ðA12:5aÞ
and
Volm ¼ Vp eo =fzð1 ðhm =zÞð1=nÞ Þ þ hm =ð1 nÞððhm =zÞð1=n 1Þ 1Þg
for n 6¼ 1 ðA12:5bÞ
A12.3 Including resistance to flow

The incompressible flow equation includes the resistance term, giving:
L=g dV=dt FV2 =ð2gÞ ¼ h z ð2:1Þ
or
Vp de=dVol Fe þ z h ¼ 0
yielding
Vp de Fe dVol þ z dVol constant Volð nÞ dVol ¼ 0 ðA12:6Þ
If it is assumed that the relationship between e and Vol follows a
similar trend as in the resistance-free case then the equation can be
written:
Vp de F=Vp fzðVolm VolÞ þ constant lnðVol=Volm Þg dVol
þ z dVol constant Volð nÞ dVol ¼ 0 for n ¼ 1 ðA12:7aÞ

and
Vp de F=Vp fzðVolm VolÞ
þ constant=ð1 nÞðVolð1 nÞ Volð1

m
nÞ
Þg dVol þ z dVol
constant Volð nÞ dVol ¼ 0 for n 6¼ 1 ðA12:7bÞ

Integrating the equation between the limits em and e and Volm and
Vol then the results are the same for all terms as for the resistance-
free case and only the resistance term need be considered.
205
For n ¼ 1 then the resistance term becomes:

ð ð
F e dVol ¼ F=Vp z ðVolm VolÞ dVol
ð
þ constant ½lnðVolÞ lnðVolm Þ dVol
¼ F=Vp fz=2ðVolm VolÞ2

þ constant½VolðlnðVolÞ 1Þ þ Volm Vol lnðVolm Þg
Writing constant ¼ Volh ¼ Volm hm as appropriate then,
ð
F e dVol ¼ F=Vp Vol2m fz=2ð1 Vol=Volm Þ2
þ hm ½Vol2 h=ðVol2m hm Þ lnðVolÞ

Vol2 h=ðVol2m hm Þ þ 1 Vol2 h=ðVol2m hm Þ lnðVolm Þg
By substituting ¼ hm =h for Vol=Volm then the equation can be
written:
ð
F e dVol ¼ F=Vp Vol2m fz=2ð1 Þ2 þ hm ½1 þ ðlnðÞ 1Þg
ðA12:8aÞ
For n 6¼ 1 the resistance term becomes:
ð ð
F e dVol ¼ F=Vp z ðVolm VolÞ dVol
ð
ð1 nÞ
þ constant=ð1 nÞ ½Vol Volð1
m
nÞ
dVol
¼ F=Vp fz=2ðVolm VolÞ2
þ constant=ð1 nÞ½Volð2 nÞ =ð2 nÞ
Volð1
m
nÞ
Vol þ Volð2
m
nÞ
ð1 nÞ=ð2 nÞg
substituting VolðnÞ
m hm ¼ constant then,
ð
F e dVol ¼ F=Vp fVol2m ðz=2Þð1 Vol=Volm Þ2
þ Vol2m hm =ð1 nÞðVol2 h=ðVol2m hm Þ=½ð1 nÞð2 nÞ

Vol=Volm 1=ð1 nÞ þ ð1 nÞ=½ð1 nÞð2 nÞÞg
206
Pressure vessels
also Vol=Volm ¼ ðhm =hÞð1=nÞ ¼ ð1=nÞ giving:

ð
F e dVol ¼ FVp Vol2m fz=2ð1 ð1=nÞ Þ2 þ hm ð1=ð2 nÞ
þ 1=ð1 nÞð1=nÞ ½1=ð2 nÞð1=n 1Þ 1Þg ðA12:8bÞ
A12.4 Complete equations

Including the resistance term the equation becomes:
For n ¼ 1:
Vp e þ Volm fzð 1Þ hm lnðÞg þ Vol2m ðÞF=Vp
fz=2ð1 Þ2 þ hm ð1 þ ½lnðÞ 1Þg ¼ 0 ðA12:9aÞ
and for n 6¼ 1:
Vp e þ Volm fz½ð1=nÞ 1 hm =ð1 nÞðð1=n 1Þ 1Þg
þ Vol2m ðÞF=Vp fz=2½1 ð1=nÞ 2 þ hm ð1=ð2 nÞ
þ 1=ð1 nÞð1=nÞ ½1=ð2 nÞð1=n 1Þ 1Þg ¼ 0 ðA12:9bÞ
A12.5 Application of the equations
A12.5.1 Maximum expanded gas volume

To calculate the maximum expanded gas volume Volm after the initial
downsurge following a pumping failure, an explicit solution can be
obtained from the equation:
aVol2m þ bVolm þ c ¼ 0
where
a ¼ þF=Vp fz=2ð1 Þ2 þ hm ð1 þ ½lnðÞ 1Þg n ¼ 1
or
a ¼ þF=Vp fz=2½1 ð1=nÞ 2 þ hm ð1=ð2 nÞ þ 1=ð1 nÞð1=nÞ
½1=ð2 nÞð1=n 1Þ 1Þg n 6¼ 1

where
b ¼ zð 1Þ hm lnðÞ n¼1
207
or
b ¼ zðð1=nÞ 1Þ hm =ð1 nÞðð1=n 1Þ 1Þ n 6¼ 1
where c ¼ Vp eo .
The value of ¼ hm =ho is obtained with ho being the initial steady
flow absolute pressure head in the vessel and z is the level difference
between the pressure vessel and the downstream reservoir þ atmos-
pheric pressure head. Note that ho ¼ z þ Feo .
Example
Z ¼ 100 m, L ¼ 1000 m, D ¼ 1:0 m, eo ¼ 0:2, F ¼ 20, ho ¼ 104 m and
hm ¼ 20 m. Vp ¼ 785 m3 and ¼ hm =ho ¼ 20=104 ¼ 0:19231
n¼1
a ¼ ð32:618343 þ 9:8128514Þ 20=785 ¼ 0:5810316
b ¼ 80:769231 þ 32:973173 ¼ 47:796058
c ¼ 785 0:2 ¼ 157
p
Volm ¼ f47:796058 ½2284:4632 4 ð0:581036Þ 157g
=½2 ð0:5810316Þ
¼ 3:163 m3
The corresponding result with zero friction is:
Volm ¼ 785 0:2=ð80:0 þ 32:188758Þ ¼ 3:284 m3
n ¼ 1:001
a ¼ ð32:592745 þ 9:807059Þ 20=785 ¼ 0:580527
b ¼ 80:737531 þ 32:968 ¼ 47:769531
c ¼ 157
p
Volm ¼ f47:769531 ½2281:9281 4 ð0:580527Þ 157g
=ð1:161054Þ
¼ 3:165 m3
n ¼ 1:2
a ¼ ð27:891309 þ 8:661267Þ 20=785 ¼ 0:4898135
208
Pressure vessels
Table A12.1. System characteristics
Project Pipe Diameter Length Velocity z Head loss hm

material (mm) (m) (m/s) (m) (hf ) (m)
Kielder Steel 2000 6 170 2.010 207.70 5.000 123.700

New Aleppo DI 1400 12 500 2.057 106.70 21.180 50.890
Kurdaha DI 400 4 803 0.955 179.59 8.320 134.000
Shiskine PE 110 2 427 0.760 38.66 12.000 24.490
Haidaria DI 1000 1 225 1.415 53.80 1.470 30.172
Silverburn DI 300 3 150 0.854 68.59 6.162 38.355
b ¼ 74:687762 þ 31:62364 ¼ 43:064122

c ¼ 157
p
Volm ¼ ½43:064122 ð1854:5186 þ 307:60288Þ=ð0:979627Þ
¼ 3:5059 m3
Volm ¼ 785 0:2=ð73:84679 þ 30:76605Þ ¼ 3:6443 m3
n ¼ 1:41
a ¼ ð23:763877 þ 7:5740078Þ 20=785 ¼ 0:4124807
b ¼ 68:940375 þ 30:004902 ¼ 38:935473
c ¼ 157
Volm ¼ ð38:935473 42:130855Þ=ð0:8249614Þ ¼ 3:873372 m3
Volm ¼ 785 0:2=ð68:064289 þ 29:111493Þ ¼ 4:031 m3
Predictions of maximum expanded air volume have been assessed for
a range of different pumping systems. The following tables contain the
characteristics of some schemes (Table A12.1) and maximum gas
volumes estimated using the present equations and obtained from
detailed elastic analysis (Table A12.2).
A12.5.2 Peak upsurge pressure head

To calculate the peak return upsurge pressure head hp in the pressure
vessel, the value of Vp e ¼ 0 since the reverse velocity has become
zero at the peak of the upsurge. Writing hm =hp ¼ the equations
209
Table A12.2. Gas volumes
Project Max. air volume, Max. air volume, Max. air volume, estimated,
detailed analysis estimated, n ¼ 0 n ¼ 0, zero friction
(m3 ) (m3 ) (m3 )
Kielder 180.00 188.000 200.600

New Aleppo 160.00 173.840 229.000
Kurdaha 4.10 3.700 4.400
Shiskine 0.10 0.140 0.225
Haidaria 14.00 14.900 15.890
Silverburn 0.73 0.873 1.040
become:
for n ¼ 1:
zð 1Þ hm lnðÞ þ ðFÞVolm =Vp ðz=2ð1 Þ2
þ hm ½1 þ ðlnðÞ 1ÞÞ ¼ 0 ¼ fnðÞ ðA12:10aÞ
and for n 6¼ 1:
zfð1=nÞ 1Þ hm =ð1 nÞðð1=n 1Þ 1Þ
þ ðFÞVolm =Vp ½z=2ð1 ð1=nÞ 2 þ hm ð1=ð2 nÞ
þ 1=ð1 nÞð1=nÞ ½1=ð2 nÞð1=n 1Þ 1Þg ¼ 0 ¼ fnðÞ

ðA12:10bÞ
It is not possible to obtain an explicit solution for and it is necessary
to resort to an iterative solution. The form of fn() is as illustrated in
Fig. A12.2. The solution for hp is the smaller of the two values of .
At the turning-point d½fnðÞ=dhp ¼ 0. For the resistance-free case
and with n ¼ 1 then,
Volm zðVol=Volm 1Þ Volm hm lnðVol=Volm Þ ¼ fnðhÞ
or in terms of hm =h and dividing throughout by Volm ,
zðhm =h 1Þ hm lnðhm =hÞ ¼ fnðhÞ
Differentiating with regard to h,
d½fnðhÞ=dh ¼ zhm ð1=h2 Þ þ hm =h
at a turning-point d½fnðhÞ=dh ¼ 0 so that,
z=h þ 1 ¼ 0 or h ¼ z
210
Pressure vessels
65 fn(x)
60
55
50
45
40
35
30 x = hm/hp
25
20
15
hmax = hm exp{z/hm}
10 x2 = hm/hmax
5 fn'{x} = (fn{x2} – fn{x1})/(x2 – x1)
–5
–10 1st approx.

to hp
hm = 40
–15
–20 x1 = hm/z hp = 372.6 hmax = 487.3

z = 100.0
–25
0 100 200 300 400 500
hp (m)
Fig. A12.2. Curve of fn() plotted against hp
The minimum value of hp ¼ hmin is set equal to z.

A maximum value of hp ¼ hmax can be obtained by assuming
hm =h 1, then,
z hm lnðhm =hÞ ¼ 0
or
lnðhm =hÞ ¼ z=hm or hmax ¼ hm expðz=hm Þ
Initial values of are then obtained from 1 ¼ hm =hmin and
2 ¼ hm =hmax and corresponding values of fnð1 Þ and fnð2 Þ found.
211
The function gradient is then approximated by:

fn0 ðÞ ¼ ffnð2 Þ fnð1Þg=f2 1 g
A new estimate of is then obtained from:
n ¼ fnðn 1 Þ=fn0 ðn 1 Þ or n ¼ n þ n 1
Example
The following values were used to illustrate the iterative solution which
can be easily programmed.
n ¼ 1, z ¼ 100 m, hm ¼ 20 m, L ¼ 1000 m, D ¼ 1 m, hmax ¼
hm expðz=hm Þ ¼ 20 expð100=20Þ ¼ 2968:3 m, hmin ¼ z ¼ 100 m,
FVolm =Vp ¼ 20:0 3=785 ¼ 0:0764
Since it is only an estimate of peak head that is required, the process
can be truncated when a sufficient degree of convergence has been
achieved.
For the Kielder project, maximum upsurge pressure in the vessels was
estimated at 373.9 m absolute. The detailed analysis gave a maximum of
335 m absolute.
A12.5.3 Required throttling

To calculate the required amount of throttling to achieve a set maximum
upsurge pressure hp in the pressure vessel the equations are rearranged to
Table A12.3. Example values
hp fnðhp Þ fn0 ðhp Þ hp
2968.3000 2.97650330 0.0171504 173.550

2794.7476 1.29153690 0.0176300 73.259
2721.4884 1.30000000 0.0181000 71.722
2649.7700 0.77560800 0.0184300 42.085
2607.6854 0.47911000 0.0186210 25.730
2581.9556 0.28840000 0.0187400 15.392
2566.5636 0.17347730 0.0188070 9.224
2557.3397 0.10427650 0.0188500 5.532
2551.8077 0.06265840 0.0188800 3.320
2548.4881 0.03764300 0.0188910 1.993
2546.4954 0.02261160 0.0188990 1.196
2545.2990 0.01358150 0.0189050 0.718
2544.5806 0.00815727 0.0189090 0.431
2544.1492 — — —
212
Pressure vessels
yield an explicit solution for F. Again setting ¼ hm =hp , then,

F ¼ fzð 1Þ hm lnðÞg=fVolm =Vp ðz=2ð1 Þ2
þ hm ½1 þ ðlnðÞ 1ÞÞg for n ¼ 1
and
F ¼ fzðð1=nÞ 1Þ hm =ð1 nÞðð1=n 1Þ 1Þg
=fVolm =Vp ðz=2ð1 ð1=nÞ Þ2 þ hm ð1=ð2 nÞ
þ 1=ð1 nÞð1=nÞ ½1=ð2 nÞð1=n 1Þ 1ÞÞg

when n 6¼ 1 ðA12:11bÞ
Example
The desired value of maximum pressure hp in the vessel is selected.
Volm ¼ 3 m3 , Vp ¼ 785 m3 , hm ¼ 20 m, hp ¼ 500 m, z ¼ 100 m.
n¼1
F ¼ f96:0 þ 64:377516g=f46:08 þ 16:624899g 785=3
¼ 280:92
n ¼ 1:001
F ¼ f95:987117 þ 64:416g=f46:067633 þ 16:629134g 785=3
¼ 280:62263
n ¼ 1:2
F ¼ f93:160096 þ 70:9976g=f43:394018 þ 17:219815g 785=3
¼ 221:56
n ¼ 1:41
F ¼ f89:801103 þ 75:596298g=f40:321191 þ 17:373275g 785=3
¼ 161:9722
The throttle at the vessels’ inlet in order to reduce the peak pressure
to 252 m absolute was estimated at 230 for n ¼ 1 while the detailed
analysis gave a throttle value of 125. Use of n > 1 gave higher values
of throttle coefficient, for example n ¼ 1:2 produced an estimated
throttle coefficient of 260.
213
A12.6 Pipeline system of varying cross-section

Where the pipeline is composed of more than one pipe of different
cross-section placed in series then it is necessary in calculations to sub-
stitute an equivalent single pipe having uniform characteristics.
The momentum of the mass of water in the equivalent pipeline must
remain the same as the momentum of water in the real system of pipes.
Length of the equivalent pipeline may be the same as the real system.
Velocity varies inversely as cross-sectional area, so that the equivalent
cross-sectional area Aeq is:
X X
Aeq ¼ Li = ðLi =Ai Þ ðA12:12Þ
214
13
Further aspects of pressure
vessels
13.1 Pressure vessel types and their fittings

A range in pressure vessel types can be found with considerable variation
in size. Vessel capacity can range from volumes of individual installations
as small as 0.1 litres to multiple vessel systems of 1000 m3 or more. These
vessels all provide some form of surge suppression function for a variety of
pipeline systems and liquids. One form of classification depends upon the
way in which the gas or air charge is maintained.
13.2 Vessels having an air—water interface

One large group of vessels comprises those in which the gas charge is in
direct contact with the liquid in the system (Fig. 13.1). The vessel
volume is split between the overlying gas charge and the stored liquid,
with transient analysis yielding the proportions of gas and liquid within
the tank. A gradual rise of liquid level may indicate absorption of gas by
the liquid or a leak in the air-filled part of the vessel. Maintenance of
an adequate gas charge within the vessel is essential if sufficient protection
is to be achieved. Many installations use an air compressor set or sets to
provide a supply of pressurised air to make up for any loss due to air
being absorbed by the liquid. For drinking water systems it is essential
to provide a high quality of purity in the air being delivered to the
vessel. In other instances a nitrogen gas charge may be used. This gas is
introduced to the vessel using cylinders of compressed gas which are
coupled to an inlet valve on the vessel.
13.2.1 Air compressors

The compressor sets should therefore be of an oil-free pattern such as
carbon or PTFE ring piston types or if standard reciprocating compressors
215
Air inlet solenoid valve Air venting valve
Bourdon-type
pressure gauge
Silicon fluid filling point
High-level alarm
Manway
Compressor start
Compressor stop
Low-level alarm
Pressure relief valve Water level indicator
Inflow/outflow
connection
Drain valve
Fig. 13.1. Typical vertical vessel arrangement
are used three-stage oil removal filters can be incorporated. These filters
will reduce oil carryover to <0.003 ppm where the temperature of air
flowing into the filter has been cooled to <308C after leaving the com-
pressor. Compressor capacity should be a balance between not allowing
the set to run too long and overheating, and too short not allowing the
set to reach a correct working temperature. Unmanned installations
should incorporate low oil level protection. Critical installations may
use duty and standby compressor sets.
While not absolutely necessary, air receivers are often installed in
surge vessel systems. Instead of the compressor delivering direct to
the vessel, a solenoid valve is included downstream of the receiver
and this opens to allow a flow of air into the vessel.
216
Further aspects of pressure vessels
13.2.2 Control of gas charge/liquid level

Several methods may be used to regulate the volume of gas within a
vessel, including:
(a) A single probe extending over the full operating range within the
vessel. The level of liquid against the probe produces a signal
proportional to the elevation of the gas—liquid interface. This
type is suitable where the liquid does not contain impurities
which might otherwise form a coating on the probe surface and
cause deterioration of the signal. Clearly this type is unsuitable
for sewage and dirty water.
(b) A set of probes of different lengths which correspond with the alarm
levels and compressor start/stop levels. Being activated purely on
contact with the liquid surface, this arrangement can be used
with sewage and other contaminated liquids.
(c) Differential pressure measurement can be used to signal changing
liquid level.
(d) Combined sight glasses and level indication, for example using a
magnetic float. As level rises a series of magnetic switches flip
over showing a change of colour, often red to indicate liquid.
With level probes and pressure measurements a sight glass should
also be included for rapid visual checking of level. In the event of
damage to the sight glass, valves at the top and bottom vessel/glass con-
nections should close automatically to prevent loss of vessel function.
Some vessel manufacturers such as Quietflo offer the option of
adding silicon-based absorption-retardant fluid to the tank. This
forms a barrier membrane on top of the liquid and reduces the rate of
gas loss. Silicon fluid use is approved by the food industry.
Liquid level within the vessel may vary considerably during a surge
event. Level may range outwith the compressor start/stop positions and
even the high and low level alarm points. In the event that a liquid
level above the compressor start level is indicated, a time delay is incor-
porated to prevent start during a transient event. This delay can be deter-
mined by the transient analysis with a default setting of say 5 min set by
the vessel supplier. Site adjustment of the delay should be possible.
The compressor should start operation or the inlet solenoid valve
should open when liquid level exceeds the compressor start level. Air
inflow to the vessel should continue until liquid level has fallen to
the compressor stop/solenoid valve close position. The compressor
should be arranged to stop when the liquid volume has been reduced
by around 5%.
217
Where a pumping system operates over a range of delivery pressures,

say high daytime pressure and lower night time pressure, the vessel gas
charge will expand as pressure drops, leaving a reduced water volume in
the vessel. To restore sufficient water volume, some air can be vented
from the vessel to ensure sufficient water is contained for surge pur-
poses. Transient analysis will yield information on the necessary
volumes of air/water over the range of operating pressure conditions.
If a large installation requires several vessels for surge protection
then, provided the vessels are all on one level, they can be treated as
one vessel controlled by a single-level sensing system. A single solenoid
valve controls air flow to all vessels, ensuring that the same pressure is
achieved throughout.
The same compressor/air receiver system may be used to supply air to
vessels serving several mains at the same pumping station. These mains
may operate at different pressures, with each main requiring its own
transient analysis. Each vessel will have its own level sensing system
and solenoid-operated air inlet valve. If a vessel is receiving air through
its valve and a second vessel requires air then the first vessel should
complete filling and its solenoid valve shut before the second vessel
starts to receive air. Ideally, two solenoid valves should be installed in
parallel at each vessel to provide backup in the event of a valve failure.
13.2.3 Other vessel fittings

To allow access for inspection and maintenance, a manway should be
included in the vessel; typically this will be around 500 mm diameter.
A drain valve should be sited at the lowest point on the vessel for
emptying.
A Bourdon pressure gauge can be installed towards the top of the
vessel connected to the gas charge.
A pressure relief valve can be included in the lower part of the vessel
so that if part of the gas charge has been lost and transient pressures
exceed the normal maximum surge level, this valve will open to prevent
further significant excess pressure rise. Some suppliers include a relief
valve on top of the vessel with air being discharged but this allows
the remaining gas charge to be further depleted.
Vessels installed out of doors may require winter protection,
especially if in an exposed location subject to wind chill when tempera-
ture can fall to 258C. Tanks and all exposed connections should be
lagged and trace heated as shown in Fig. 12.19. This also applies to
sight glasses, as freezing can cause failure of the tube. System design
218
should consider site conditions carefully and all failure conditions which
might develop.
A circulation system can be included in those larger installations
containing a substantial volume of water in the vessel. Eventually the
water can become stale unless an arrangement to change the water is
incorporated. A small pump and pipework system providing about
5 litres/min (7 m3 /day) will ensure that water in the vessel remains
sweet.
13.3 Bladder vessels

An alternative type of vessel uses an elastomer bladder, usually
neoprene, contained within the vessel. A valve is connected to the
top of the vessel to allow recharging with gas. This can be done using
a portable cylinder containing either nitrogen or compressed air.
Response of a vessel of this type can be analysed in much the same
way as a conventional vessel and will be subject to some of the same
uncertainties as the conventional vessel analysis such as the validity
of the gas law used.
While a considerable variation in design is evident in bladder-type
vessels, they have the common feature of having a membrane
separating the liquid and gas components. In addition to finding
application as straightforward transient protection devices, many
bladder-type installations are used as pulsation dampers upstream and
downstream of positive displacement pumps and to compensate for
temperature changes within closed pipeline systems. By providing
complete separation of gas and liquid, the possibility of gas charge
loss through absorption is eliminated and with it the requirement for
a compressor and level monitoring system. The life expectancy of the
bladder may be considered a possible drawback.
Figure 13.2 illustrates three forms of bladder vessel. Figure 13.2a is a
fairly typical arrangement of vessel available for surge suppression duty.
The bladder contains the liquid volume which does not come into
contact with the vessel walls. This has some benefit in reducing main-
tenance of the inside of the vessel. A nitrogen gas charge fills the space
between bladder and vessel wall. A gas filling connection is provided on
top of the vessel with filling achieved using a compressed nitrogen
bottle. Vessel volumes available as standard may range from 8 to
50 000 litres.
Figure 13.2b illustrates a smaller type of accumulator. In this case the
bladder contains the nitrogen charge with the liquid filling the space
219
Gas filling connection
Air or nitrogen
Liquid-filled bladder
Outlet to pipeline
(a)
Protective cap Non-return

gas valve assembly
Steel shell Nitrogen
Separator Gas charging valve

bladder
PTFE diaphragm
Poppet valve
Nitrogen
Anti-extrusion ring
Shell in stainless steel,
polypropylene and PVC Unrestricted
Liquid inlet/outlet inlet–outlet port
(b) (c)
Fig. 13.2. Bladder vessel types
between the bladder and the vessel walls. As pressure drops and the gas
expands, then to prevent possible expansion of the bladder out of the
vessel a poppet valve is included which is pushed shut by the expanding
bladder thus preventing the bladder being damaged. Its chief function is
to store fluid under pressure. Stored fluid is then released to the pipeline
circuit to supplement pump output and to absorb pressure peaks and
shocks generated elsewhere in the system.
220
Figure 13.2c shows a small diaphragm pulsation damper. It comprises

a dished shell in two halves. The upper half contains a gas charging
valve and the lower the inlet/outlet connection. Between these
halves is clamped a diaphragm, possibly of PTFE. Bonded in the
centre of the circular diaphragm is a ‘button’. In the event of pressure
drop the gas expands and the diaphragm moves downwards. The
button will eventually make contact with the inlet/outlet port,
preventing further movement and avoiding damage to the diaphragm.
This small vessel is used as a pulsation damper and typical volumes
may range from 0.1 to 30 litres.
13.4 Positioning a pressure vessel

Siting of pressure vessels is important in a number of respects. Consider
the layout of pressure vessels at Riding Mill pumping station as shown in
Fig. 13.3. The vessels are offset from the main by about 30 m. Following
pump trip, a rarefaction pressure wave from the pumping station
travels to the vessel connection. The wave will split with part conti-
nuing along the main while a second component propagates along
the vessel connection. Only when this second component has been
reflected from the vessels in the form of a compression wave, and
returns to the main, will the vessels start to influence events in the
Air vessel chamber

Pumping station building
45.0 m
21.
4m
3 No. 403 mm orifice

plates in series in 813 mm
inlet bypass line
Flowmeter 2220 mm o.d.
Stilling well chamber
outlet chamber m
25
57.
1820 mm o.d.
Air vessel 3.5 m o.d. ¥ 11.0 m

w
Flo
Sight Max WL
glass
Pumping main Min WL
2000 mm o.d. Connection for

future pumping main
Section of air vessel chamber
Fig. 13.3. Pressure vessels at Riding Mill
221
Anti-vacuum valve
Triple air valve cluster
Varies
Typical arrangement of
anti-vacuum valve Letch House tank
250 Elevation = 223 mAOD
200 Penn’s Hill

Elevation (mAOD)
River Tyne and

150 Riding Mill PS
100 Air vessels
50
16.5 m
P
0
0 930 1400 1760 2160 2455 3050 3815 4230 4865 6207
Chainage (m)
Fig. 13.4. Rising main pipeline profile
pipeline. During the time delay of around 0.06 s before the vessels are
able to respond, the first component of wave has been travelling
along the rising main shown in Fig. 13.4. The resultant effect is a
short-duration pressure pulse which travels back and forth in the
rising main on top of the main surge controlled by the pressure ves-
sels. Should this short-term pressure fluctuation coincide with the
peak or trough of the primary wave, the effect will be to increase or
decrease the maximum or minimum transient pressure based upon
the main wave only. The predicted and observed head variation at
Riding Mill pumping station following simultaneous trip of six duty
pumps, can be seen in Fig. 13.5. Irregularities in head variation are
present on both traces.
Within the Riding Mill pumping main, this effect was responsible for
operation of an air valve at Penn’s Hill (Fig. 13.6), although a
preliminary analysis which neglected the effect of vessel connection
length had suggested that no such valve operation would occur. Both
the observed and simulated head variations at Penn’s Hill show clearly
the presence of higher-frequency components of wave motion. These
waves travel together with the primary wave which is dominated by
the presence of pressure vessels. Where possible, the pressure vessels
should be placed as close as possible to the pumping main to avoid
unnecessary delay in vessel response and to minimise the development
of these secondary wave components.
222
320
6 pump trip record at Riding Mill PS
300
d
ve
280
er
ed
bs
260 ict
O
240 ed
Pr
220
Elevation (mAOD)
200
180
160
140
120
100
80
60
40 n = 1.4
20 a = 900 m/s
0 K = 85
0 10 20 30 40 50 60 70 80 90 100 110 120 130 140 150

Time (s)
Fig. 13.5. Head variations at Riding Mill — 6 pump trip
A good-sized connection to the vessel is also desirable to avoid

excessive head loss during outflow thus ensuring the most effective
response from the air chamber. It may be desirable to install a ‘swept’
tee to minimise head loss as flow enters the rising main.
For maximum effect a pressure vessel should be placed as low as
possible. Where static head is low then the difference in level from
?60
Recording failure 6 pump trip record at Penn’s Hill
t ed
e dic
?40 Pr
?20
Elevation (mAOD)
Observe
?00
d
?80 Air valve elevation
Air valve/anti-vacuum valve operation n = 1.4

?60 a = 900 m/s
K = 85
Pipeline elevation
0 10 20 30 40 50 60 70 80 90 100 110 120

Time (s)
Fig. 13.6. Head variations at Penn’s Hill — 6 pump trip
223
the pipeline to the centreline of the vessel may be quite significant. The
smaller the difference in level from the vessel centreline to the down-
stream discharge level the larger will be the required vessel volume.
In systems of low static it is important to employ a realistic assessment
of vessel elevations if a possible underestimation of vessel volume is to
be avoided.
It has been said that the function of the pressure vessel is to limit
changing head or pressure along the rising main. This leaves the rela-
tively short stretch of pipeline between the suction well upstream of
the pump and the connection with the pressure vessel to be considered.
The function of the pressure vessel is to sustain pressure at the vessel
connection and along the rising main. However, on the downstream
side of a failing pump, pressure can fall relatively quickly and a steep
adverse hydraulic gradient can be established within this relatively
short section of pipeline between the pump and the vessel connection
where pressure is being maintained at a relatively high level (Fig. 12.4).
Compared with the situation without a pressure vessel, jdV=dtj
downstream of the vessel is reduced while jdV=dtj upstream of the
vessel is increased.
Flow will tend to reverse more quickly upstream of the vessel connec-
tion than if the vessel were not included. This has important implica-
tions with respect to closure of the check valve downstream of the
pump and will be considered in greater detail when discussing check
valves in Chapters 20 and 21. It may be noted that siting of the
vessel close to the pumps will tend to aggravate check valve closure
conditions and if practical the vessel should be moved further down-
stream rather then being say teed off a manifold directly opposite the
pump branch connections (Fig. 13.7).
Suction well
Pumps
To rising main
Check valves
Discharge manifold
Pressure vessel
Preferred vessel location Less desirable vessel locations
Fig. 13.7. Pressure vessel locations
224
The pressure vessel does not suppress the transient at its source as
does the flywheel, but rather the transient is developed at the pump
and the vessel acts to suppress its effects to an acceptable extent as
the pressure wave enters a downstream pipeline for instance. As
described, the positioning of the vessel and configuration of its con-
nection can influence the effectiveness of response.
13.5 Installation with air valves

Figure 13.8 shows an hybrid form of vessel incorporating some of
the attributes of a conventional installation but with an air valve
arrangement as a means of replenishing the air charge. In a typical
example the chamber is fitted with a ‘dip-tube’ which extends down-
wards from the air valve connection on top of the vessel, to about
the mid-height of the tank. The lower end of the tube is open. Figure
13.9 illustrates the normal sequence of operation of this protection
device when installed downstream of a pumping station.
(a) At static, the air valve is closed and the tube is filled with liquid. A
charge of air is enclosed in the upper part of the vessel above the
lower end of the tube and under the static pressure head.
(b) While pumping, the air valve remains closed with the tube liquid-
filled as before. The air charge is compressed under the pumping
head in the top of the vessel.
(c) When head starts to fall after a pumping failure, the air mass starts
to expand, releasing liquid into the rising main thus suppressing the
transient head drop in the main. Thus this vessel acts in the same
manner as a conventional pressure vessel at this stage. The dip-tube
still remains liquid-filled at this time.
(d) After a time when head has fallen to the air valve operating level,
the valve will open to allow air to flow into the tube and finally into
the vessel itself. This air flow augments the expanded air volume
already within the vessel. Operation of the air valve largely stabi-
lises head at the vessel and the continued flow of liquid out of
the chamber is essentially balanced by air entering the tank via
the air valve.
(e) After flow reversal has occurred, head starts to rise at the vessel
with air flowing out of the air valve and liquid entering the
vessel. When liquid level reaches the lower end of the dip-tube
the air mass contained in the vessel outside the tube is isolated
while the air in the tube exits via the air valve. The air valve
225
Protective cap
Pressure gauge
Air regulating valve Central tube
Flanged outlet
Interchangeable pipeline
connection DN 100 to DN 400
To
pipeline
Fig. 13.8. Vessel with air valve and central tube
shuts when liquid level reaches the top of the tube and the air mass
remaining in the vessel is compressed by the return flow in the
vessel.
As the hydraulic transient decays, conditions are gradually restored

to static as in item (a) above; the air charge has been replenished as
necessary by operation of the air valve. For this type of vessel to
226
Steady pumping head
Static head
Air valve (AV) closed

AV closed
Air charge in vessel
Rising main Flow from pump
(a) (b)
Head shortly after trip
AV closed
Outflow from vessel
(c)
Head rising during reversed flow
Head approaching minimum

Air inflow at AV Air outflow at AV
Outflow from vessel Inflow to vessel

(d) (e)
Fig. 13.9. Sequence of operation of a vessel with air valve
function successfully, the transient piezometric level must fall to allow

air valve operation. The pipeline profile must be suitable to allow this to
happen. A vessel of this form provides protection from the moment of
pumping failure and relies on the air valve at a later stage both to
augment the air mass and to top-up the air charge contained in the
chamber. In principle a vessel of this type could be installed at any
appropriate location along a pipeline where hydraulic conditions are
suitable, as it does not rely on a power supply. Some throttling of
227
Centrifugal pumping station
Surge suppression tanks
Twin GRP rising mains
General layout
Air valves Air valves
Upper tanks fitted

with air valves Mild steel air valve
support frame
Lower tanks replenished
with air during downsurge
after pump trip
Concrete tank supports
Vessel connection lines
Elevation through surge suppression tanks
To rising mains
Fig. 13.10. Alternative vessel arrangement using air valves
inflow to the vessel might also be considered, as in a conventional vessel

connection. Capacity of the vessel should be sufficient to contain the
maximum air volume during the sequence of operation.
The example shown in Fig. 13.8 can occur in different forms as
illustrated in Fig. 13.10. Twin pumping mains are each fitted with two
horizontally aligned vessels. The upper vessel has three sewage air
valves installed along the top of the horizontal vessel. The lower vessel
is connected to the upper vessel at the base of the riser, extending upwards
to the higher of the vessels (Fig. 13.11). During pumping, the hydraulic
gradient ensures that air is purged from the upper vessel with the air
valves closing on completion of air venting. The air charge in the lower
vessel is compressed but with its air mass remaining in the chamber.
228
Fig. 13.11. Vessels with air valves
In the event of pump shutdown, pressure falls and the air mass in the
lower chamber expands allowing liquid to flow from the vessel thus
providing a measure of surge protection. Only when hydraulic gradient
has fallen to the operating level of the air valves will these valves open
to allow air inflow which augments surge protection.
229
14
Surge tanks and related
structures
These devices are chambers usually open to the atmosphere, although

pressurised versions have been known. Analysis of a pressurised surge
tank follows the same process as that for a pressure vessel. The simplest
type of surge tank will be a vertical cylinder either above ground
or below ground as at Strathblane on the Loch Lomond Scheme
(Fig. 14.1). There will also be a connection to the pipeline or tunnel
system. Often surge tanks will be of considerable capacity, particularly
where these are used to suppress transients occurring in hydroelectric
power systems. Small surge chambers are also found in water supply
systems where there is a need to control pressure fluctuations.
14.1 Purpose of a surge tank

As with all forms of surge alleviation, the surge chamber is intended to
reduce the rates of change of velocity which would otherwise occur if
the chamber were not present. The presence of a free water surface has
the effect of producing a strong reflection of a pressure wave much
like the reflection from a reservoir. By providing a source of water and a
means of storing water, the tank is a pressure-relieving mechanism. In
the case of flow deceleration in a pipeline downstream of the surge tank
connection, water from an upstream pipeline will be diverted into the
chamber thus avoiding the need for the same rate of flow deceleration
in the upstream line. Likewise, when flow is accelerating in the down-
stream line the chamber supplies water to the downstream main avoiding
the requirement for the upstream line to supply flow to the downstream
system at the same rate. By reducing the rates of acceleration and
deceleration in the upstream conduit, pressure surging upstream of
the surge tank connection can be correspondingly lessened.
230
Surge tanks and related structures
Cap
Ground level
M
TWL = 76.81 mAOD

(pumps at Ross Priory tripped)
Surge shaft at north portal

of Balmore tunnel
2134 j tunnel
1524 j steel pumping main
Pumping level
M
M
M
Pumping flow
Tunnel water level maintained
by throttle valve
Reversed flow in main while downstream = 67.67 mAOD
pressure vessels and
Static level in shaft = 68.58 mAOD
feeder tanks refilling
Fig. 14.1. Simple surge shaft between pumping main and gravity flow tunnel
There are many instances where a tank or chamber is included,

whose function is not that of a surge tank but whose response to and
effect upon hydraulic transients can be analysed using the same tech-
niques as for a surge chamber.
The suitability of a site for this type of installation depends upon the
level of the maximum piezometric line relative to the ground surface —
that is, maximum water level in the chamber is coincident with piezo-
metric level.
Analysis often has to take account of complexities of shape, especially
where the chamber is below ground, often including galleries or a sur-
face pond in place of an upper gallery as at Clachan HEP station
(Fig. 14.2). Galleries, both upper and lower, effectively concentrate
the chamber volume at these levels where it will be most effective
rather in the manner of an RSJ section where area is concentrated at
a distance from the axis. By having a more modest shaft section, the
water level will rise and fall in the shaft more rapidly than if the area
of chamber was constant with changing water level. This means that
an effective hydraulic gradient necessary for acceleration and decelera-
tion of flow is attained more quickly with a smaller shaft cross-section.
231
Dam crest = 304.49 mAOD Inflow from Clachan burn
305 M
Surge pond
300 Intake and screens

cill = 302.36 mAOD
Inclined surge shaft
295
M
290 Surge gallery
Low pressure aqueduct

M
Elevation = 285.0 mAOD

285
Steel-lined penstock to
underground generating station
Fig. 14.2. Underground surge chamber installation
Having reached the gallery elevation, water level is largely stabilised

with an effective hydraulic gradient in the system to produce the
required acceleration or deceleration of flow.
Emptying and filling of galleries during transient events can introduce
further complications to analysis, in the form of surface waves in the
galleries as described by Ellis and Khairulla (1974). Control sections
can be introduced at the outfall from a gallery or at the connection
from a headpond as at Clachan HEP station (Fig. 14.2).
14.2 Simple analysis

The simplest model of a surge tank might include the following ele-
ments (Fig. 14.3). Conservation of volume within the tank dictates
232
Piezometric level in
pipeline at connection
Piezometric level
at water surface M
Ht H
Vt At
+
Ac
Au/s Ad/s
Vu/s Vd/s
+ +
Fig. 14.3. Definition sketch for surge shaft analysis
that:
Vu=s Au=s At dHt =dt ¼ Vd=s Ad=s ð14:1Þ
Head difference between the connection and the water surface
Ht H is made up of two components: inertial head and head loss in
the connection, so that:
Ht H ¼ fKL þ 1gfAt =Ac dHt =dtg2 =ð2gÞ þ Lt =g dVt =dt ð14:2Þ
where KL is the overall loss in the connection, At is cross-section of the
chamber, Ac is cross-sectional area of flow in the connection and
Vt ¼ dHt =dt.
In addition the Cþ and C characteristics provide the following rela-
tionships at the time when the wave paths arrive at the tank connection:
Vu=s þ g=au=s H ¼ Jþ and Vd=s g=ad=s H ¼ J
Four equations are available to solve for the unknowns, Vu=s , Vd=s , H
and Ht . The above equations can be solved in a straightforward manner
with dHt =dt ðHt Hto Þ=t.
14.3 Long connection to a chamber

Surge chambers can be large and in consequence there may be a signif-
icant time lag in their response. The above equation for conservation of
water volume in the tank assumes that there will be an immediate effect
233
upon water level in the chamber in the event of a transient event

arriving at the chamber connection. If the distance along the axis of
the chamber from the free surface to the connection is Lt , then the pres-
sure wave reflection time will be 2Lt =a. As already indicated in respect
to offsetting pressure vessels from a pipeline (Chapter 13), this can
allow time for part of a transient to bypass the connection before
there is a response from the chamber. This effect will be more significant
when the connection is of modest cross-sectional area Ac and the
connection is relatively long. The effect will also be more significant
when the area of the surge chamber is modest and galleries provide
much of the volume. Response time will also vary with changing
water level in the surge chamber with the time for a pressure wave to
be reflected back from the free surface at its greatest when water
level is at its maximum.
Surge chambers have also been constructed using sections of pipe.
One example of this is at the Pao La Balsa scheme in Venezuela
where a tee was formed on the pumping main and a branch pipeline
laid on the hillside (Fig. 14.4). Sections of pipe may be laid on a
rising slope to provide a surge shaft. Because of the length of this
branch it is best modelled as an actual pipeline of variable length.
The pipeline is schematised in the usual way with head Hj at the con-
nection to the rising main represented by the equation:
X .X
Hj ¼ 1=g ðA JÞ ðA=aÞ ð6:4Þ
Since cross-sectional areas are the same and wave speed is also
assumed common, this reduces to:
X
Hj ¼ a=g ðJÞ=3 ð14:3Þ
At the free surface (Fig. 14.4), assuming the positive direction of flow to
be from the shaft into the pipeline, then for conservation of volume:
VAp ¼ As dHs =dt ð14:4Þ
or averaging over a time increment t:
ðV þ Vo ÞAp =2 ¼ As ðHs Hso Þ=t
giving:
V þ Vo ¼ 2As =ðAp tÞðHs Hso Þ
Also from the characteristic arriving at the free surface:
V g=aHs ¼ J
234
Piezometric level at
the connection point
Piezometric level
at the free surface
Ap
Hj Ht
Dx
Ap
Dx +
Ap
Dx
Fig. 14.4. Inclined pipeline as a surge shaft
This neglects the inertia of that part of the water column above the first
computational section. Substituting for V in the conservation equation
at the free surface:
Hs ¼ ð2As =ðAp tÞHso J Vo Þ=ð2As =ðAp tÞ þ g=aÞ ð14:5Þ
Having found Hs , the velocity at the free surface can be obtained from:
V ¼ J þ g=aHs
In this way the response of the shaft is included and the effect of a
pressure transient partially bypassing the shaft can be modelled. Varying
water level in the shaft requires some method of identifying those
sections of shaft which are above water level and only the parts of
235
the system below the water level included in the analysis. Where raw
or treated water is being handled there is also the possibility of allow-
ing some spillage from the top of the surge pipeline provided the loss
of water is acceptable and the spillage can be conveyed safely to a
convenient watercourse.
Dawson (c. 1980) described use of pipes as a form of surge tank in
sewage pumping applications. Clearly any spillage from the top of the
surge pipe has to be avoided and there is some risk of offensive
odours being vented when sewage level in the shaft is rising. Sizing of
the surge shaft requires consideration of possible septic conditions
developing if the sewage is allowed to dwell within the shaft for too
great a time. During downsurge the shaft must essentially empty.
Thus the shaft cannot have too great a capacity. On the other hand,
if too small a shaft volume is provided it may not fulfil its surge suppres-
sion function and violent oscillations may develop which could
adversely affect other parts of the system such as check valves.
14.4 Full-size connection

The choice of method of analysis is dependent upon the characteristics
of the shaft and connection. In the simple arrangement shown in
Fig. 14.5 where the connection is the full size of the surge chamber
then the ratio of head at the connection given by the equation:
X X
Hj ¼ 1=g ðJAÞ= ðA=aÞ ð6:4Þ
to head Hv ¼ aVo =g at a downstream shut valve can be expressed as:
Hj =Hv ¼ 2A=ð2A þ As Þ
assuming a constant wavespeed. If As A then Hj will be small and a
simplified analysis which treats the connection to the surge tank as a
reservoir of constant water level may be sufficient. This was the tradi-
tional approach in the days before use of computer programs became
widespread. Where the connection is of modest size and appreciable
length it may be appropriate to use a more detailed approach.
14.5 Extent of protection

The surge tank does not offer 100% protection against pressure waves
arriving at the tank connection. Again, consider the simple arrange-
ment depicted in Fig. 14.5. The surge chamber is connected to a
pipeline such that the upstream and downstream sections of pipeline
236
Inertia head rise

after valve closure
dx/dt = V – a
Steady flow piezometric level Surge shaft Hi = a/gVo
As Valve
Ls
A A
V = Vo V = 0.0
Upstream pipeline length Downstream pipeline length

Lu/s/a >> Ls/as Ld/s/a >> Ls/as
Fig. 14.5. Simple shaft with pressure wave
have wave propagation times much longer than in the surge tank itself
— that is, Lu=s =a Ls =as and Ld=s =a Ls =as . If a valve at the down-
stream end of the pipeline is abruptly closed with an initial steady
flow velocity Vo the inertial head rise Hi will be aVo =g as shown. The
pressure wave thus created will travel upstream to the surge chamber
connection at a rate dx=dt ¼ V a. At the connection the head rise
which occurs as the wave front arrives is given by:
Hj ¼ fVo A ðVo ÞA þ 0As g=f2A þ As g
For simplicity, if a common wavespeed a is assumed for both pipeline
and surge shaft then:
Hj ¼ a=gVo 2A=ð2A þ As Þ ð14:6Þ
This head rise will induce a velocity Vs into the surge chamber given by:
Jþs ¼ 0 ¼ Vs þ g=aHj
or
Vs ¼ 0 g=aHj ¼ g=aa=gVo 2A=ð2A þ As Þ
¼ Vo f2A=ð2A þ As Þg ð14:7Þ
and an invariant value
Js ¼ Vs g=aHj ¼ 2Vo f2A=ð2A þ As Þg ð14:8Þ
237
Table 14.1. Ratio of head rise after reflection to initial head rise at connection
Time from arrival Ratio of inertial head rise Hjn =Hj1 at junction
of pressure wave
0 1
2Ls =a 1 A
4Ls =a 1 2A þ A2
6Ls =a 1 3A þ 3A2 A3
8Ls =a 1 4A þ 6A2 4A3 þ A4
10Ls =a 1 5A þ 10A2 10A3 þ 5A4 A5
12Ls =a 1 6A þ 15A2 20A3 þ 15A4 6A5 þ A6
At the water surface in the surge chamber it is assumed in the short

term that no significant change in water level has occurred so that head
remains constant at Ho ¼ 0. Velocity at the free surface is then:
V ¼ Js ð14:9Þ
and the invariant travelling from the free surface back to the surge shaft
connection is:
Jþs ¼ V ¼ Js ¼ 2Vo f2A=ð2A þ As Þg
At the connection after time 2Ls =a inertia head Hj is now given by:
Hj2 ¼ a=gð2Vo A 2Vo As f2A=ð2A þ As ÞgÞ=ð2A þ As Þ
¼ a=gVo 2A=ð2A þ As Þ½1 2As =ð2A þ As Þ ð14:10Þ
Setting, 2As =ð2A þ As Þ ¼ A the ratio of head rise Hj2 =Hj1 ¼ 1 A .
After subsequent wave reflections at intervals of 2Ls =a, values of Hj are
given by the sequence in Table 14.1.
The response of the surge shaft varies with the ratio As =A and
Table 14.2 illustrates the influence of this ratio on the magnitude of
head rise at the shaft connection.
Table 14.2. Head rise at the shaft connection as a proportion of the head rise aVo =g
Time (2Ls =a As =A ¼ 0:5 As =A ¼ 1:0 As =A ¼ 2:0 As =A ¼ 4:0 As =A ¼ 8:0

increments)
0 0.800 0.67000 0.5 0.33300 0.2000

1 0.480 0.22100 0 0.11100 0.1200
2 0.288 0.07300 0 0.03700 0.0720
3 0.173 0.02410 0 0.01230 0.0430
4 0.104 0.00800 0 0.00410 0.0260
5 0.062 0.00260 0 0.00140 0.0156
6 0.037 0.00087 0 0.00045 0.0093
238
Pressure wave travelling upstream

from surge tank connection
2Ls intervals Pressure wave reflected

from surge tank connection
travelling downstream
Inertial head rise = aVo/g
Surge shaft
Upstream pipeline Downstream pipeline
Fig. 14.6. Pressure wave reflections from a surge shaft
Due to the delay in surge shaft response, a pressure wave component

is able to travel into the upstream tunnel. As wave reflections
return from the free surface in the shaft to the connection with the
pipeline, relief of pressure occurs. The resulting compression wave
propagating upstream from the shaft connection has a form deter-
mined by the ratio As =A as shown in Table 14.2. Figure 14.6
illustrates conditions subsequent to arrival of the initial pressure wave
from downstream.
14.6 Other aspects

Throttle arrangements can be employed in the connection to a surge
shaft in the same manner as at the inlet to a pressure vessel (Fig. 12.16).
Where the surge shaft is part of a system in which governing is
employed to maintain either a constant power output or a constant
flow rate, the stability of oscillations within the surge shaft should be
investigated. Jaeger (1977) describes in detail both the development
of stability criteria and their application.
Complicated surge chambers may involve a mixture of pressurised
flow coupled with some elements of free surface motion for example
(Fig. 14.2).
239
14.7 Initial estimates of surge tank parameters

Both graphical and analytical techniques have been developed to assist
the engineer in determining preliminary design parameters.
Fok (1980) prepared a quite comprehensive set of design charts for
surge tanks placed downstream of a pumping station on the discharge
line. His graphical approach allows pipeline resistance and the effects
of a differential throttle to be included subsequent to a power failure
at the pumping station. Peak transient piezometric level and minimum
level can be obtained together with their times of occurrence. Also the
maximum reversed flow rate can be obtained and the necessary time
delay before pumps may be restarted.
Considering a surge tank placed upstream of a hydropower station,
Jaeger (1977) presented some interesting analytical results for surge
tank behaviour following a turbine trip. He attributes the following
equations to work by Prášil and Eydoux (as cited in Jaeger, 1977). Start-
ing from the dynamic equation of mass oscillations in a pipeline or
tunnel, Prášil produced an equation for the highest water level in the
surge chamber. This equation was incapable of direct solution and
required use of tables. By making a simplifying assumption, Eydoux
was able to produce equation (14.11) which is capable of direct
solution.
zmax ¼ 1 2=3Fro þ 1=9F2ro ð14:11Þ
where
p
zmax ¼ z=z ; z ¼ Vo fLAt=ðgAs Þg and Fro ¼ FVo2 =z
where z is measured from an upstream reservoir level as datum.
Equation (14.11) gives essentially the same result as the equation of
Prášil for values of Fro 0:7. While intended for use with a surge
tank upstream of a turbine house, the same equation is equally applic-
able to a booster pumping station. It may be possible to use the same
analytical approach to produce an equivalent equation for a surge
tank downstream of a pumping station.
14.8 Related structures

Many instances exist of a chamber having a water surface in contact
with the atmosphere. Under transient conditions the water level in
such tanks will vary over time. The ability of the chamber to store
water or to act as a local reservoir for water can have beneficial effects
with respect to transient events. Range in tank water level requires to
240
be established, both to prevent the chamber from overtopping and to

avoid air ingestion at low water levels. Usually range in water level
can be limited by increasing the surface area of a chamber but in the
case of sewage installations this may not be attractive as an enlarged
area of tank may provide greater opportunity for settlement and a
longer dwell time for sewage. Containment of maximum water levels
can be handled by providing an adequate freeboard. Some cases of
such chambers will be considered.
14.8.1 Service reservoir as a one-way surge tank

The pumping station and suction reservoirs of a large water distribution
scheme are shown in Fig. 14.7. Three rising mains, each DN 1200, are
connected to a surge vessel installation. In addition a bypass line, also
DN 1200, extends from reservoir No. 1 to a valve chamber from
which connections are made to the rising mains. The pressure vessels
provide initial surge suppression in the event of pumping failure but
should piezometric level in the rising mains at the connection fall
6 5
4 3
2 off DN 1600 suction mains
Pumping station
2 1 Surge
vessels
NRV
DN 1200 bypass
3 off DN 1200 rising mains
Fig. 14.7. Service reservoir and bypass line
241
below the prevailing water level in reservoir No. 1 then a check valve on
the bypass line will open to permit flow from reservoir No. 1 into the
rising mains thus augmenting flow from the vessels and inhibiting
further substantial head drop in the mains. The bypass line is some
60 m in length and overall losses in this line should be taken into
account in any simulation.
14.8.2 Operation of an existing service reservoir

The City of Harare water supply system contains a large number of
service reservoirs (SRs) supplied via five large trunk mains. These
mains are filled by pumping treated water from Warren Park Pumping
Station (Fig. 14.8). When pumps are idle, reticulation flows can
occur from some of these reservoirs which are thus equipped to allow
both inflow and outflow. To protect the ring mains following a pumping
failure at Warren Park, involving simultaneous failure of 12 duty
pumps, a set of pressure vessels was recommended with a gross
volume of 550 m3 . During the downsurge following trip of pumps,
head in rising main No. 1 falls below the water level in the nearby
975 mm
800 mm
750 mm
300 mm
Warren Park PS
375 mm
1000 mm
1500 mm
Viking SR complex
450 mm
Fig. 14.8. Service reservoirs as surge chambers
242
Viking SR, causing outflow from this storage facility. This outflow pro-
vides a source of water for surge suppression in addition to that from the
pressure vessels. The vessels’ volume could be reduced by 50 m3 were
Viking SR relied upon to yield water after all pump trip events.
Use of service reservoirs in this way depends upon the existence of a
backflow connection. Otherwise, instead of outflow from the reservoir,
air valves may open allowing the line to deprime. If there are no air
valves, high vacuum pressures may occur in the branch serving the
reservoir. Even if it is not the intention to utilise an SR for surge
duties in this way, hydraulic transients are such that the reservoir
branch will be affected and should therefore be included in any
modelling exercise. If it is not the intention to allow reversed flow
from the SR and it is necessary to avoid vacuum pressures then a
check valve can be included on the branch main close to its connection
to the trunk main.
14.8.3 Filtration plant

First consider the Three Valleys Water North Mymms Ultrafiltration
membrane plant shown in Fig. 14.9. This installation removes particles
down to the macromolecular level with a relatively modest head drop
through the membranes themselves. At a design flow rate of
1590 m3 /h, eight membrane units will be in service at any time, with
a further unit possibly undergoing programmed maintenance.
This system obtains water from a feed tank just upstream of the plant
building itself and delivers treated permeate to a contact tank a
relatively short distance outside the plant (Fig. 14.10). Since water
levels in the feed and contact tanks vary only by a maximum of
200 mm, the static head is very small in comparison with head losses
associated with system components.
The membrane units of Fig. 14.9 are at a height of around 3 m above
the normal water level in the feed tank which is at 78.5 mAOD. The
absence of any effective static lift means that parts of the membrane
plant are subject to vacuum pressures both under steady flow and
also static conditions. Figure 14.11 shows a section through the pipeline
system together with the static and steady flow gradients.
To ensure that there is no risk to water quality in the underground
pipeline from the membrane plant to the contact tank, positive
pressures should be maintained along this stretch following pumping
failure. Figure 14.11 also shows maximum and minimum hydraulic
gradients after trip of two duty pumps. Vacuum pressures were predicted
243
244
Primary units
Secondary
units
EL +81.0
M
GL +78.80
M GL +78.70
CIP M
f600
f600
f700
EL +75.50
M
Feed pumps Backwash pumps Primary Dirty backwash

strainers Lifting davit
holding tank 40 m3
Section A–A
Fig. 14.9. Membrane units

Contact tank
WL = 78.5 mAOD
DN 450 DN 600
DN 500
DN 600
Location of surge tank

Feed tank
WL = 78.5 mAOD
UF membrane plant
Permeate line to
contact tanks
Fig. 14.10. Membrane plant layout
along the downstream pipeline. A small chamber, surface area 3.14 m2 ,

was installed on the downstream side of the membrane plant to alleviate
transient conditions along this pipeline after pumping failure. By
reducing the rate of flow deceleration dV=dt in the pipeline, þve
pressures were maintained (Fig. 14.11). Peak water levels in the tank
were also determined during pump startup to ensure that adequate
freeboard was provided.
14.8.4 Seawater intake system

A second example concerns the seawater intake system shown in Fig.
14.12. Twin 2 m diameter pipelines of average length 226 m lead
from intakes to inlet chambers of a pumping station. Each inlet pipe
245
Steady pumping – 2 pumps

Piezometric line – max. resistance
Head drop through membrane unit
Steady pumping
Piezometric line –
min. resistance Steady flow – max. resistance
min. resistance
Primary strainers
Surge tank Min. piezometric level
Membrane with surge tank
units
Feed tank Contact tank
WL 78.5 WL 78.5
Within membrane plant building Underground pipeline
Sub-atmospheric head
Min. piezometric level without protection
Fig. 14.11. Head changes through membrane plant
supplies two pumps from a single stilling chamber. A maximum of three

pumps will operate at any time — that is, two pumps drawing from one
intake and the third pump from the second intake. The water level in a
stilling chamber is a function of the prevailing sea level and the number
of pumps operating. Maximum level will occur if two pumps drawing
from the same intake are tripped together when sea level is at its max-
imum. Minimum level will be established when two pumps are started
in the same stilling chamber with minimum sea level.
Since change of water level in the chamber is small during one
time step it is reasonable to calculate conditions of flow through
pumps using a suction water level held constant at the initial level
for that increment. Thus flow rate Q through pumps can be found
and averaged over the increment. The equation for conservation of
volume gives:
AðV þ Vo Þ=2 ðQ þ Qo Þ=2 ¼ At dH=dt ð14:12Þ
246
Crane 30 T
+13.50
Trash rack Stilling

Stop chamber Pipes Cables
gate Stop Drum screen Stop reserve
Forebay chamber reserve
Seawater Overflow +5.50 gate gate
intake weir Cooling
+2.04 HWL 5158 water
+1.04 MWL Max. tidal water = +1.13 pumps
+0.04 LWL MSL = 0.00
–0.70
Min. tidal water = –1.40
–4.00
Inlet
–6.50 pipes –6.50
–8.5
–9.50
2000
5200 13004 4800 4500 4200
Fig. 14.12. Seawater intake system

247
0.048
6.166
12.285
18.403
24.521
30.640
36.758
42.876
48.994
55.113
61.231
67.349
73.467
79.585
85.703
91.822
97.940
104.058
110.176
116.294
122.412
128.531
134.650
140.769
146.888
153.007
159.126
165.245
171.365
177.484
183.603
189.722
Level (mASL) –0.5
–1.0
–1.5
–2.0
–2.5
Time (s)
Fig. 14.13. Head variations at seawater intake chambers
and the Cþ characteristic yields:

V þ g=aH ¼ Jþ
Unknowns V and H can be easily found.
As an illustration consider the case where three pumps are operating
and that a single pump using one of the stilling chambers is tripped. The
remaining two pumps operating together with the second stilling cham-
ber continue to function. Figure 14.13 shows the predicted changes in
water level in the two stilling chambers. Water level in the chamber of
the pump which has been tripped increases towards the prevailing sea
level while water level in the other chamber falls to a lower level as
the operating pumps deliver increased flow. This increased flow
comes about as a consequence of falling head in the common down-
stream discharge header after the pump failure.
14.8.5 Seal weir

At the downstream extremity of a cooling water circuit, the pipeline
system is often maintained at a predetermined head by providing a
seal weir (Fig. 14.14).
When tank head H > Zw flow over the weir is given by:
p
Q ¼ Cd B ð2gÞðH Zw Þ3=2 ð14:13Þ
and when H Zw then:
Q ¼ 0:0
248
Qw
Seal weir
H
Inlet pipeline
Zw
Cross-section = A
Tailwater
Fig. 14.14. Seal weir arrangement
Variations of coefficient of discharge Cd for the weir can be found

from the literature for different weir forms and tabulated as a function
of head over the crest H Zw for interpolation during computations.
The initial value of crest head Ho Zw is usually adequate to give a
value of Cd for each time increment.
For conservation of volume in the tank:
AðV þ Vo Þ=2 ðQw þ Qwo Þ=2 ¼ At dH=dt
and
dH=dt ¼ ðH Ho Þ=t
and from the Cþ characteristic:
V þ g=aH ¼ Jþ
The equations can be easily solved for V and H.
Figure 14.15 shows a typical example of changing tank level after trip
of three operating pumps.
14.8.6 Water towers

Other structures may be specifically constructed either to improve
normal operational behaviour but with some benefits to transient
conditions, or may be created specifically for the purpose of alleviating
transient conditions.
249
5.4
5.3
5.2
Level (mASL)
5.1
5.0
4.9
4.8
0.048
6.166
12.285
18.403
24.521
30.640
36.758
42.876
48.994
55.113
61.231
67.349
73.467
79.585
85.703
91.822
97.940
104.058
110.176
116.294
122.412
128.531
134.650
140.769
146.888
153.007
159.126
165.245
171.365
177.484
183.603
189.722
Time (s)
Fig. 14.15. Head variation upstream of seal weir following pumping failure
An example of a structure which improves normal operational

behaviour is a water tower (Fig. 14.16). A structure such as this
would be constructed in an area where topography does not provide a
sufficiently elevated site for a local ground storage tank. The plan
arrangement of a number of such towers is shown in Fig. 14.17. The
Geassa elevated tank (water tower),

Eastern Liwa, Abu Dhabi, UAE
M 175.000 mASL TWL
M 171.650 mASL BWL
M 153.700 mASL GL
Fig. 14.16. Typical water tower
250
Al Raiqa ELT
Mizaira’a
ND 400 DI
ND 150
Mizaira’a PS
Chlorination
Surge protection
To Nafeer
ND 400
ND 150
ND 150 DI
Sabkha ELT
ND 400
ND 400 DI
ND 150 DI
Gurmeda ELT
ND 150 DI
ND 400 DI
Shah ELT
Tharwaniya GT ND 150 DI
ND 400 DI
ND 400 DI
ND 150 DI
Nashash ELT ND 150 DI

BPT
To Hameem
Hawatheen ELT
Fig. 14.17. Pipeline system serving water towers and ground tanks
251
210
190
170
Elevation (mOD)
150
130
110
4 8 12 16 20 24 28
Chainage (km)
Fig. 14.18. Envelope curves for pump trip with water towers connected
tower provides a storage volume at a level adequate to provide a

continuing gravity flow to downstream consumers when inflow to the
water tower is interrupted. Flow conditions in the upstream pipeline
supplying the tower are improved by lifting the hydraulic gradient to
a higher level. Following a pumping failure, with the towers installed,
minimum transient head conditions are more likely to remain positive
(Fig. 14.18) than without the towers in place (Fig. 14.19). The
210
190
170
Elevation (mOD)
150
130
110
4 8 12 16 20 24 28
Chainage (km)
Fig. 14.19. Envelope curves for pump trip without water towers
252
elevation of the tower also produces a steeper deceleration gradient

within the upstream pipeline, allowing the water column to come to
rest more quickly. This will assist in limiting the required capacity of
other surge devices such as an air vessel at the pumping station.
14.8.7 Special structures

Many outfall systems consist of relatively flat landward sections of
pipeline followed by seaward pipelines which terminate in diffusers.
Consider the example shown in Fig. 14.20. Twin pipelines were of
Pumping station
Pipe bridge
Static hydraulic
gradient
Air valve
Maximum flow
Piezometric level
hydraulic gradient
after pump trip
Twin 534 j GRP

landward pipelines
Reduced flow
hydraulic gradient
Loop at seawall
Reduced flow hydraulic

gradient at HWL
Maximum flow
Hydraulic gradient at
HWL,
Twin seaward pipelines
LWL
Static hydraulic gradient at

HWL,
LWL
LWL HWL
Diffuser section
Fig. 14.20. Effluent outfall system
253
19.xxx m
8 No. 305 × 127mm × 37 kg

1 m UBs
25 thick open mesh galvanised flooring 4 No. 5126 mm long
14.276 m 4 No. 5980 mm long
2971
Detail P
i.l. 13.450 m Drg No. P2325/5/R16
Detail D 2 No. 152 × 152 mm × 23 kg
Drg. No. P2325/5/R16 1 m UCs
MS galvanised access ladder

with rectangular safety cage
Detail No. 1
2971
4 No. 365 × 368 mm × 177 kg
Galvanised 76.2 × 3.25 mm dia 1 m UCs
Weldmesh (Ref. no. 310) supported (4 No. 13.362 mm long)
on angles to within 1829 mm of Note Column ends fabricated
base: thereafter mesh is flush for bearing
25.4 × 25.5 × 2.64 mm dia.
(ref. no. 1118) 20 No. 150 × 90 mm angles
Detail F
8.476 m Drg. No. P2325/5/R15
2971
Detail G
See detail L
Drg No/ P2325/5/R15 Drg No. P2325/5/R16
Inner tie 25 dia.
305×127 mm×37 kg m
UB (2 No. 5632 mm long)
16 No. 60 × 60 mm angles
(angles back to back)
2432
Detail No. 1
4.355 m
534 mm dia. flange plain ended
GRP pipes 1000 mm long
I.L. 3.720 m
2000
2445
Scour chamber, floors and 3.500 m
1829
walls lined with G.R.P. 2.300 m 3.200 m
200
200
D 2.520 m 2.526 GL. 2.50 m

Detail K Drg No.
P2325/5/R15 2.026
1178
Ion
I.L. 1.450
1.100 m
Blass infill
200
200
215 1500 215

1930 See detail J
Drg No. P2325/5/R15
2 No. 534 m dia. G.R.P. pipes
concreted in to base and
upstand
Fig. 14.21. Inverted U pipework arrangement with air valves
534 mm diameter thin-walled glass reinforced plastic (GRP). Only quite

small sub-atmospheric pressures were allowable.
A structure which was specifically constructed to improve transient
conditions is shown in Figs 14.21 and 14.22. This inverted U arrange-
ment was formed to improve minimum transient head conditions along
the effluent pipelines discharging through an outfall beneath a tidal
estuary. At the top of the loop of pipework is a set of three sewage air
valves. The pipeline profile is illustrated in Fig. 14.20. During pumping
at normal maximum flow rate, the upper parts of the loop of pipework at
254
Fig. 14.22. Inverted U showing air valves and vessels
the seawall are primed with the air valves shut. Hydraulic gradient over
the air valves is sufficient to keep the valves tight shut. At lesser flow
rates the air valves will be open and the top of the loop represents
the downstream limit of the pumping mains. At low flows and/or low
tide level, the seaward section of the system flows under gravity. This
arrangement has the merits of providing an almost constant static
head against which pumps can operate. As far as pressure transient
behaviour is concerned, the loop of pipework creates a useful adverse
hydraulic gradient along the landward stretch of pipeline, which
produces a more rapid flow deceleration along this part of the system.
Sub-atmospheric pressures are thus minimised and quantities of air
admitted through operation of air valves along the landward stretches
of pipeline are reduced. Demands on pressure vessels located at the
pumping station are also more modest, allowing smaller vessels to be
installed. Without the inverted U of pipework at the seawall, large sec-
tions of the landward pipelines would deprime after a pumping failure,
more substantial vacuum pressures would occur and the size of pressure
vessels would have to be greatly increased. The resulting process of
restarting pumps would involve removal of large volumes of air from
the system before a steady pumping regime could be established.
255
Piezometric level
Air inflow
Air valve
(closed) Zu/s Zd/s
–
M –
M
Hu/s = Hd/s
Pressure
vessel Hu/s
Hd/s
Vu/s Vu/s
Vd/s Vd/s
(a) (b)
Air inflow Air inflow
Zu/s
M
Zd/s
M
Hu/s Zu/s
M
Hd/s
Zd/s
Hu/s M
Vu/s Vu/s Hd/s

Vd/s Vd/s
(c) (d)
Air outflow Air outflow
M
–
M
Hu/s Hd/s
Hu/s
M
Vu/s Hd/s Vu/s
Vd/s Vd/s
(e) (f)
Fig. 14.23. Sequence of operation of inverted U arrangement
Modelling hydraulic conditions at the loop of pipework requires

consideration of a number of regimes, just some of which are depicted
in Fig. 14.23. After each time increment of a simulation the following
variables must be established.
256
Upstream and downstream piezometric level Hu=s and Hd=s ;

upstream and downstream velocity Vu=s and Vd=s ;
upstream and downstream water level Zu=s and Zd=s ;
air inflow/outflow rate through air valves Qair ; and
gauge pressure head within the air mass h.
Equations available are as follows.
The quasi-invariant relationships from upstream and downstream
characteristics:
Vu=s þ g=aHu=s ¼ Cþ and Vd=s þ g=aHd=s
Air flow relationship for the air valves:
Qair ¼ function ðhÞ
fobtained from valve supplier, inflow assumed þ veg
Conservation of volume within the air mass:
ðVd=s Vu=s ÞA Qair ¼ dðVolÞ=dt
Vu=s A ¼ dZu=s =dt and Vd=s A ¼ dZd=s =dt
Polytropic relationship for the air mass:
ðh þ hatm ÞVoln ¼ constant fover a time stepg
Also
Hu=s Zu=s ¼ h ¼ Hd=s Zd=s
Not all of these equations are applicable for each flow regime of
Fig. 14.23.
(a) During steady pumping at maximum design rate and with higher
tide levels the loop is fully primed with piezometric level above the
operating level of air valves. Air volume in the downstream pressure
vessel is compressed to its minimum at this stage. Vu=s ¼ Vd=s ,
Hu=s ¼ Hd=s , Qair ¼ 0:0 and Vol ¼ 0:0 at this time. Solution is
achieved using the quasi-invariant relationships alone.
(b) After pumping failure, piezometric level at the loop falls and air
valves open. Air inflow commences and an air pocket develops at
the top of the loop. While the air—effluent interface lies above
pipe invert level at the top of the loop then Hu=s ¼ Hd=s ,
Vu=s < Vd=s and h is ve. Air mass in the pressure vessel starts to
expand.
257
(c) Air inflow continues and effluent flows over the top of the loop
cascading into the downstream vertical pipe with level falling in
this part of the loop. Air charge within the pressure vessel continues
to expand. Hu=s < Hd=s , Zd=s < Zu=s , Vu=s < Vd=s , Qair is þve and h
is ve.
(d) Effluent level in the downstream vertical section falls below the
vessel connection and the air masses are united. Upstream flow
will finally reverse as effluent moves upstream attempting to refill
the vessel at the pumping station and removing air admitted to
the pipeline through upstream air valves. The effluent level falls
in the vertical upstream leg of the loop. Eventually upstream and
downstream effluent levels are stabilised with the seaward level
equal to the prevailing tide level. Pressure of the air mass becomes
atmospheric.
(e) When pumps are restarted effluent level in the upstream leg rises
and pressure in the air mass his þ ve. The increase in pressure
starts to push air out of the air valves on top of the loop and
volume of the air mass starts to decrease. Increased pressure
causes a small positive velocity to develop in the downstream
pipeline.
(f ) With continued pumping the upstream leg is filled and flow
cascades into the downstream pipe producing a level increase and
isolating the vessel. Under the action of rising downsteam level,
flow develops in the seaward section of pipelines. If flow is below
the maximum or if tide level is low, the loop may remain only
partially primed. When an equilibrium condition is achieved and
Vu=s ¼ Vd=s , air venting ceases with the remaining air mass being
at atmospheric pressure h ¼ 0:0.
258
15
Feeder tanks or volumetric
tanks
Some pipeline profiles are ‘unfavourable’ with respect to avoiding

negative pressures (Fig. 15.1). In the case of a treated water main it
would not be acceptable to permit minimum pressures to fall to a
point where negative pressures developed in the pipeline. Providing a
pressure vessel arrangement of sufficient capacity to alleviate unaccep-
table minimum pressures to an adequate extent throughout the system
may not be a practical proposition because of the very large vessel
capacity required.
It is a general principle that the requirement for pressure transient
protection can be most efficiently met by installing the equipment as
close as possible to the area affected. Additional protection can be
installed at an intermediate summit or a high point on the pipeline.
For a treated water pipeline an additional pressure vessel might be
installed at the summit. Alternatively this local protection may take
the form of a surge tank. If the height of a surge chamber was imprac-
tical or visually unacceptable then a feeder or volumetric tank can be
used. Like a pressure vessel, a feeder tank acts as an alternative
source of water following a pumping failure. By supplying water to a
downstream pipeline, rates of deceleration are reduced and with
these corresponding pressure changes.
15.1 Components and location of a feeder tank

The primary elements making up a feeder tank are illustrated in Fig. 15.2.
Being open to the atmosphere a feeder tank will behave in some
respects as a surge tank. The main difference is that the feeder is
isolated from the pipeline under steady flow and only comes into
operation during a transient event when piezometric level falls below
259
Service
reservoir
Minimum hydraulic gradient following M
a pumping failure and with pressure
vessel at PS for protection
Pipeline profile
Minimum piezometric level following a

pumping failure and without protection
Possible location
of feeder tank
Pressure vessel
Pumping station
Fig. 15.1. Pipeline showing possible feeder tank location
Float-operated valve
TWL
Feeder tank
Filling
connection BWL
Non-return valve (NRV)

outflow only
Isolating valve
Upstream pipeline
Isolating valve Downstream pipeline
Fig. 15.2. Schematic of feeder tank showing principal features
260
Feeder tanks or volumetric tanks
the water level in this tank. Feeder tanks will usually be located at a
high point on the main where preliminary analysis without local protec-
tion has shown transient sub-atmospheric pressures to occur.
To limit maximum water level in the feeder tank a non-return valve
(NRV) is positioned in the outlet connection. This NRV shuts as flow
attempts to re-enter the tank, forcing flow through the relatively
modest filling connection. As water level in the tank rises, an inlet
valve, often a float valve, progressively closes, shutting off inflow as
level approaches TWL. The filling connection is sized so as to restrict
inflow rates to ensure that no secondary surging of any significance is
generated during filling. Capacity of the feeder tank should be arranged
so that no air is able to enter the pipeline.
15.2 Mode of operation

Assuming the feeder tank to be filled at the outset, then supposing a
pumping failure occurs at a pumping station towards the upstream
end of the main. A fall in pressure or downsurge, will travel downstream
from the pumping station. The decreasing piezometric level is accom-
panied by a reducing upstream velocity and deceleration dV=dt.
While the declining piezometric level remains above the water level
within the feeder tank, the NRV will remain closed and the tank will
have no influence on the passage of the transient along the pipeline.
Deceleration dV=dt will be the same downstream of the tank connec-
tion as in the upstream main.
It is advisable when developing a piece of software to consider abnor-
mal circumstances and to ensure that the modelling process can cater
for all eventualities. In the present context this would include, among
other scenarios, prediction of tank emptying and development of an
air pocket in the pipeline (Fig. 15.3b—e). Such abnormal behaviour
has been observed and will be discussed later in this chapter. A brief
description of some of these configurations is given below.
(a) Only when piezometric level in the main at the tank connection
has fallen below the prevailing water level in the tank, does the
NRV open to allow outflow from the tank into the pipeline. At
this stage:
Vu=s Au=s þ Q ¼ Vd=s Ad=s ð15:1Þ
that is, the downstream flow is augmented by outflow from the tank
thus reducing the rate of deceleration of flow in the downstream
pipeline below that in the upstream part of the main — that is
261
NRV open NRV open

Hu/s
Hd/s
Hu/s
Hd/s
(a) (b)
NRV open NRV open
M
M M
M
Hu/s Hd/s Hu/s

Hd/s
(c) (d)
Hu/s Hd/s
NRV closed NRV closed
Hd/s
Hu/s
(e) (f)
Fig. 15.3. Operating sequence of feeder tank
jdV=dtd=s j < jdV=dtu=s j. Corresponding changes in transient pres-

sure are also reduced and piezometric level at the tank connection
is determined by the relationship:
nX o
Htank Hmain ¼ KL þ 1 Q2 =ð2gA2c Þ ð15:2Þ
262
where
P Ac is the cross-sectional area of the outflow connection and
KL is the sum of loss coefficients for the components making up
this connection — that is, entry loss, NRV loss, tee, isolating valve
and pipe resistance. The feeder tank supplies treated water from the
tank to the main, thus minimising risk of contamination.
Characteristics are selected which arrive at the feeder tank
connection at the time when a solution is required. The quasi-
invariant values propagating along these paths yield a pair of
equations at the feeder connection so that:
Vu=s þ g=aHmain ¼ Jþ and Vd=s g=aHmain ¼ J
These equations can be solved for Vu=s , Vd=s , Hmain and Htank for
normal operation.
(b) After the feeder tank and its connection have drained, initially a
common water surface and piezometric level will exist upstream
and downstream of the feeder connection. Conservation of
volume is given by:
Vu=s A þ Qair ¼ Vd=s A þ As dz=dt
where As is the air—water interface area in the pipe and
z ¼ zu=s ¼ zd=s is the water surface level.
(c) If the downstream water level should fall then a situation may
develop where upstream and downstream water levels become
different so that:
Vu=s A ¼ Asu=s dzu=s =dt þ q
and
Vd=s A ¼ q Asd=s dzd=s =dt
where q is the spillage from upstream to downstream.
(d) If a pressure vessel at an upstream pumping station is refilling, velo-
city upstream of the feeder connection may reverse, causing the
water level to decline on this side of the feeder connection, then:
q ¼ 0:0
(e) If downstream flow should reverse, spillage may occur from down-
stream to upstream of the feeder connection, so that:
Vu=s A ¼ Asu=s dzu=s =dt þ q
and
Vd=s A ¼ q Asd=s dzd=s =dt
263
(f ) During refilling water enters the feeder tank via the modest-
diameter filling connection. Inflow is regulated by a valve which
closes in response to rising water level in the tank. As level
increases, the valve progressively closes so that a quiet shut-off of
inflow is achieved as the tank level approaches top water level
(TWL).
15.3 Abnormal behaviour

A case involving abnormal feeder tank behaviour occurred after pump
trip in a treated water pumping system comprising a prestressed con-
crete rising main, diameter 1070 mm and length almost 16 km. As
can be seen in Fig. 15.4, the main has an unfortunate profile as regards
avoiding sub-atmospheric pressures. To provide adequate protection at
190
Downstream reservoir
Feeder tank location
M M
180
170
Elevation (mAOD)
160
150
Pipeline profile
140
130
1070 mm diameter prestressed concrete pipeline
Stanton steel cylinder type – 15 796 m long
M Pressure vessels
Pumping station
0 2 4 6 8 10 12 14 16 18
Chainage (km)
Fig. 15.4. Feeder tank on a rising main with pressure vessel
264
the pumping station in the form of pressure vessels alone, necessary

capacity would have been extremely large. A feeder tank of capacity
75.1 m3 , TWL of 183.354 mAOD and tank outlet level of
180.495 mAOD was provided at chainage 5.4 km. Surface area of the
tank was 26.267 m2 . The outlet connection had a diameter 381 mm
and a filling connection diameter of 127 mm.
Figure 15.5 shows the predicted variation of piezometric level at the
feeder connection for a simulated pump start/stop operation. A 16 mgd
(million gallons/day), (73 Mld) pump was started and run until steady
flow was attained after around 4 min when the pump was tripped.
The feeder tank came into operation 30 s after trip, emptying after
150 s. This prediction was confirmed by field observations. A peak air
volume of 4.52 m3 was calculated to develop in the pipeline 210 s
after pump trip. After flow reversal in the downstream section of
pipeline, the check valve on the outflow connection closed and the
air pocket was expelled through the filling connection. When water
started to pass through the filling connection the sudden increase in
flow resistance caused a modest head rise to occur and a deceleration
of the reversed flow in the downstream pipeline.
The pressure vessels at the pumping station reached their maximum
expanded air volume and started to refill about 90 s after pump trip.
Until this time the feeder tank had only been supplying water to
maintain positive flow in the downstream pipeline. When the pressure
vessels started to refill, the feeder tank provided the most convenient
supply of water for the vessels and so for a time this tank supplied water
both to the downstream pipeline and also to the upstream pipeline for
vessel refilling. The high demand during this time is responsible for the
rapid drawdown in feeder level. The tank outlet diameter is relatively
modest for this high rate of outflow and substantial head loss was
experienced leading to a sub-atmospheric pressure head of 4.0 mWG.
If the feeder tank forms a part of an overall protection package which
includes a pressure vessel installation at an upstream pumping station,
then particular attention should be paid to the implications of vessel
refilling on capacity of tank required and sizing of the outlet connection
for the tank.
The demand for water can be high when a feeder is required to supply
water to both upstream and downstream parts of the system and head
loss in the tank connection may be appreciable. Piezometric level in the
main below the tank will be given by:
nX o
Hmain ¼ Htank KL þ 1 Q2 =ð2gA2c Þ
265
Feeder tank
refilling
Peak of first upsurge

at the pumping station
4.52 m3 of air
admitted to
pipeline during
Feeder tank and
this interval
connection empty
Crown of Invert of pumping

pumping main at main at feeder
feeder connection connection
Feeder tank
emptying Minimum piezometric
level reached at the
pumping station
(end of first downsurge)
Trip of ‘WEIR 16’
Feeder tank
full
1 min.
Time lag till pressure

wave reaches feeder site
10 m
Start of ‘WEIR 16’
and Hmain must remain above the minimum acceptable þve pressure
otherwise a vacuum pressure may appear in the main. Note that
Htank may be at a fairly low level at this stage. In an installation
involving large pressure vessels, the demand for refilling water can be
266
substantial even where inflow to the vessels is regulated using a throttle

arrangement. If it is possible that a scheme may be augmented in the
future and that additional vessel capacity may be added and/or the
extent of flow regulation at the vessels’ connection altered, then it is
essential that the tank and its connection be sized to accommodate
these future conditions.
When the original feeder tank was designed, computer modelling was
not available and analyses were carried out in two parts using graphical
methods. First, hydraulic transients were studied in the pipeline from
the pumping station to the feeder tank site, with the feeder being
treated as a reservoir. These calculations yielded the necessary vessel
capacity. Second, the pipeline from the feeder to the downstream
reservoirs was examined to establish the necessary capacity of feeder
to control rates of flow deceleration in the downstream pipeline. It
may be that the volume of water required for refilling pressure vessels
was not added to the feeder volume obtained from analysis of the down-
stream pipeline. This could have led to the undersized installation.
Following more recent computer analysis of the complete system, a
second feeder of capacity 232 m3 was constructed with a larger outlet
connection DN 800. The enlarged outlet considerably reduced head
loss, thus improving minimum pressures in the pipeline and the
increased capacity of feeder ensured that the tank did not empty
completely even under worst-case pump failure conditions.
Where more than one form of protection is being used, in this case
pressure vessels at the pumping station and a feeder tank along the
pipeline, the configuration of one protection measure influences the
response of the other protection equipment. The capacity of pressure
vessels is considerably reduced by the provision of the feeder tank
while the feeder tank volume is partly determined by the vessel capa-
city. The extent of any throttling at the vessel(s) inlet, controls to
an appreciable extent the maximum rate of vessel refilling. If it were
necessary to reduce head loss during outflow from the feeder while
vessels were refilling, this could be assisted by introducing additional
throttling at the vessel inlet.
15.4 Mains duplication: Example 1

A second example of feeder tank application concerns a treated water
transmission system involving duplication of a rising main running
between a large pumping station and receiving reservoirs. The steel
mains each have a diameter 1524 mm and length 18.7 km. The original
267
160
Trip of 1st pump
Predicted head variation
150
Trip of 2nd pump
140
Trip of 3rd pump
Elevation (mAOD)
130
120
110
Observed head variation
100
90
2 4 6 8 10 12 14
Time (min.)
Fig. 15.6. Observed and predicted head at pumping station
rising main had a route which was quite favourable, with pressure
vessels at the pumping station alone able to avoid development of
sub-atmospheric pressures. The new main had a less attractive profile
and using only pressure vessels would require a substantially increased
capacity than the straightforward duplication of vessel volume envisaged
for the new main.
Before embarking on prediction of transient behaviour within the
duplicated system, the opportunity was taken to compare transient
predictions for the original main with field observations. This is very
desirable for all systems but often only possible on larger installations.
Figure 15.6 shows comparisons of transient piezometric level at the
pumping station for a normal auto-sequence shutdown of three duty
pumps.
Simply adding two new vessels to the existing pair was not sufficient
to avoid sub-atmospheric pressures at higher points on the main follow-
ing a pumping failure. Figure 15.7 depicts the maximum and minimum
transient head along the new main together with the initial profile of
the pipeline.
From the initial studies, possible sites for feeder tanks were identified at
high points towards the downstream end of the new main. A number of
feeder arrangements were considered with the option of using feeders at
two different sites being investigated. Since the route had not been fina-
lised, the opportunity was taken to explore alternative possible profiles.
Figure 15.8 shows one such route with two feeder tanks at chainages
14.5 km and 18.2 km. The feeder capacities for this configuration were
268
160
140
Maximum transient
piezometric level
120
Elevation (mAOD)
100
80 Minimum transient
piezometric level
60
Pipeline profile
40
2 4 6 8 10 12 14 16 18
Chainage (km)
Fig. 15.7. Envelope curve following pump failure without feeder tank
quite large and unattractive from an economic standpoint. Using two

tanks was not an attractive option and further adjustments to the
route were made in an attempt to find a more favourable profile.
Figure 15.9 shows the final choice with a single small feeder at
chainage 15.2 km.
A dramatic reduction in feeder capacity was achieved through
alteration of the pipeline profile.
160
140
Maximum transient
piezometric level Feeder tank location
120
Elevation (mAOD)
100
80 Pipeline
profile
piezometric level Feeder tank
location
40
2 4 6 8 10 12 14 16 18
Chainage (km)
Fig. 15.8. Envelope curve following pump failure with possible feeder tanks
269
160
140 Maximum transient

piezometric level
Elevation (mAOD)
piezometric level
100
80 Modified
pipeline
profile
60 Feeder tank
location
40
2 4 6 8 10 12 14 16 18
Chainage (km)
Fig. 15.9. Envelope curves following pump failure with amended profile and
feeder tank
With a feeder tank at chainage 18.2 km the hydraulic gradient

between the tank and the downstream reservoirs was relatively shallow
with only a very gradual flow deceleration occurring. While the feeder is
still supplying water to the downstream pipeline, the vessels start to
refill and this second demand has to be met from the feeder tank, result-
ing in a large capacity chamber.
Altering the route to avoid the summit at chainage 18.2 km allows a
single feeder to be placed at chainage 15.2 km. The transient piezo-
metric gradient between this single tank and the downstream reservoirs
is steeper, allowing a more rapid flow deceleration to develop with flow
reversing in this section of main before the vessels start to refill. The
water for refilling then comes from the downstream reservoirs and
not from the feeder tank. The demand on the feeder thus remains
small, giving a modest feeder capacity.
15.5 Mains duplication: Example 2

A further example of feeder tank behaviour concerns a proposal to
duplicate a raw water pumping main some 17 km in length. The steel
pipelines are 60 in. (1524 mm) in diameter. Profile of the original pipe-
line contains two important summits in the first 4 km from the pumping
station. Pressure vessels at the pumping station would be unrealistically
large to provide complete protection of the pipelines and so two feeder
270
80
60 Ross Loan feeder

Elevation (mAOD)
Minimum piezometric level

40
Gartinbantrick
feeder Pipeline
profile
20
4 8 12 16
Chainage (km)
Fig. 15.10. Envelope curves following pump failure with two feeder tanks
tanks were constructed, one at each summit. Maximum and minimum

transient hydraulic levels are shown in Fig. 15.10 following a pumping
failure with the feeder tanks in operation.
The arrangement of the newer of the two tanks at Ross Loan is shown
in Fig. 15.11. Figure 15.12 shows the main structure of reinforced
concrete under construction, while Fig. 15.13 depicts the branch
from the rising main and the isolating butterfly valve just inside the
valve chamber.
Operation of the feeder tanks was monitored and comparisons made
with predicted behaviour. For the newer tank, Fig. 15.14 shows the
rapid drawdown in water level in this tank after a pumping failure.
This was followed by a gradual recovery of water level controlled by
the throttling action of the filling connection.
The older feeder showed more irregular behaviour with free-surface
wave motion occurring in the tank both at the start of drawdown and
when refilling commenced (Fig. 15.15). These wave actions were
attributed to delayed movement of the non-return valve in the tank
outlet connection.
Detail of feeder tank level at the start of refilling is shown in
Fig. 15.16. It was evident that after the wave motion in the tank sub-
sided, an appreciable rise in water level had taken place. This indicated
that the check valve had not closed promptly after flow reversal but had
stuck in the open position, allowing inflow to the tank through the
outflow connection before the valve eventually closed. Eventual
271
272
54.600 Fall (See detail) 54.775 Fall 54.600 54.725
54.400 54.300
Item 6 Soffit 54.400
D15.
53.527
280
53.435 TWL 53.400 2
D11.
3
1
Item C
3
1118
D14. For details of access
Open
3
70 300 Wall thickening around
2621
platform see Record C.L. 51.900
400 550 600

puddle pipe
51.330 DRG. No. 7
300
T.O.C. H. wheel 500
25 mm grout
under baseplate 1000 50.200 m
D13.
49.200 49/915 D24

49.430
T.O.C. T.O.C.
48.854
300
600 mm square pipe Varies (45.630–48.480)

support block 48.239
48.129 IL
75 mm thick grade
D18.
C25 concrete as 900 ¥ 500 ¥ 1600 ¥ 800 mm ¥ 1600 ¥ 400 mm valve
blinding 390 mm high 810 mm high pipe support block
value support support block
block Visqueen D.P.M.
Fig. 15.11. Elevation of feeder tank

Fig. 15.12. Feeder tank chamber under construction
Fig. 15.13. Inlet to feeder tank under construction
273
53.6
53.4
Predicted level
53.2
Level (mAOD)
53.0
Observed level
52.8
52.6
52.4
52.2
52.0
0.335
31.825
63.315
94.805
126.295
157.785
189.276
220.767
252.257
283.747
315.236
346.725
378.214
409.703
441.193
472.682
504.171
535.662
567.154
598.646
630.138
661.631
693.123
724.615
756.107
787.599
819.091
850.583
882.075
913.567
945.059
Time (s)
Fig. 15.14. Feeder tank water level after pump trip

1m
Trip of three pumps
1 min.
Trip of two pumps
Fig. 15.15. Unsatisfactory feeder tank behaviour after pump trip
274
Trip of three pumps
Short-term level change in feeder

before NRV closure
Trip of two pumps
Short-term level change in

feeder before NRV closure
Fig. 15.16. Detail of feeder tank water level at commencement of refilling
check valve closure occurred relatively abruptly, causing a sudden

reduction of flow into the tank as flow was then forced through the
smaller filling connection. This sudden flow deceleration would produce
additional surge effects in the pipeline.
Feeder tanks, by their nature, are located some distance from pump-
ing stations but still require regular inspection and maintenance if they
are to fulfil their function. When work is being carried out on a feeder
tank, the isolating valve will be closed and hydraulic transient investi-
gations should be carried out to establish a safe pumping rate while the
tank is out of service.
A further notable effect of the feeder tanks in this installation was to
influence the peak heads at the pumping station. Figure 15.17 shows
predicted piezometric level at the pumping station after pumps were
tripped. The first upsurge after flow reversal as the vessels refill is
quite modest. At this stage the feeder tanks are supplying water to
the vessels, and head at this end of the pipeline system is largely
275
90
Steady pumping level
Elevation (mOD) 80
70 First upsurge
controlled by feeder Second upsurge while
the feeder is re-filling
60
50
1 2 3 4 5 6 7 8
Time (min.)
Fig. 15.17. Head variations at pumping station following pumping failure
controlled by the water level in the feeder and by the inertia of the
relatively short length of main between the pumping station and the
feeder tanks. During the second and much higher upsurge, the feeder
tanks are refilling and the head at their connections is much greater.
Also the inertia of the entire length of main is involved, leading to
the much greater head rise.
15.6 Aspects of feeder behaviour to consider

It is important that any simulation be of sufficient duration that the
complete picture of transient behaviour is established.
The capacity of feeder tank necessary can be quite sensitive to the
pipeline route and it may be worthwhile to consider some alternative
pipeline profiles in order to reduce the required capacity of feeder.
One of the examples discussed considered alternative alignments and
the resulting effects upon volumetric tank volume.
Refilling of the feeder is usually accomplished through reversed flow
in the pipeline system downstream of the tank although on occasion it
may be necessary to rely upon restarting of pumping to complete the
refilling process. Filling is accomplished through a smaller-diameter
connection usually fitted with a valve which closes in response to
rising water level in the tank. The check valve on the outflow connec-
tion closes as flow starts to re-enter the tank, forcing water through the
more constricted filling connection which discharges above top water
276
level. Flow resistance acts to throttle inflow thus limiting magnitude of

reversed flow and assisting to prevent secondary pressure transient
effects developing when inflow ceases. The inlet valve shuts progres-
sively as tank water level rises, with flow decelerating gradually to
avoid occurrence of secondary transients. Figure 15.14 shows a record
of correct feeder tank water level behaviour during both outflow and
inflow.
Feeder tanks are usually installed at some distance from a pumping
station and not subject to daily inspection. A prolonged interval
between inspections carries the risk of malfunction. Figure 15.15
shows a recording of water level in a malfunctioning feeder tank
during outflow and inflow. Surface waves were recorded at the start
of outflow and when flow began to re-enter the tank. These waves
were the consequence of a delay in response of the outflow check valve.
15.7 Preliminary estimation of feeder tank volume

There are two aspects to establishing an initial estimate of the required
chamber volume. Continued flow into a downstream pipeline when the
feeder has come into operation will draw water from the tank. If an
upstream pumping station is equipped with a pressure vessel for
instance then this vessel will require to be refilled following the down-
surge which results from a pumping failure. Vessel refilling occurs after
flow reversal in the pipeline between the pumping station and the
feeder tank. Depending upon relative rates of deceleration of flow
within the pipelines upstream and downstream of the feeder tank,
vessel refilling may commence while the feeder is still supplying water
to the downstream pipeline or alternatively vessel refilling may not
start until flow reversal has taken place in the downstream pipeline.
In the first instance the feeder is required to supply water both to the
upstream and downstream pipelines. Vessel refilling water comes from
the feeder in this case. If flow has reversed in the downstream pipeline
before the vessel starts to refill then refilling water comes from the
downstream pipeline to some extent.
Equation (17.9a), developed for estimation of buffer tank volume,
may also be used to estimate the volume of a feeder tank.
Volmax =Volp ¼ D=ð fLÞ lnf1 þ hf=zg ð17:9bÞ
In this case Volmax is the volume of feeder tank water supplied to the
downstream pipeline and Volp is volume of the downstream pipeline.
D, L and f are respectively the diameter, length and overall friction
277
factor of the downstream pipeline. The friction factor includes an

allowance for losses in the outflow connection of the feeder. hf is
the pipeline resistance loss during steady flow along the downstream
pipeline and z is the level difference between the feeder tank and the
discharge level downstream.
When it is considered necessary to include vessel refilling volume,
the equations derived in the Appendix to Chapter 12 could be used
or alternatively any of the graphical methods. As an example, equation
(A12.5a) might be used. For simplicity, this equation ignores the effect
of pipeline resistance.
Volm =Volp ¼ Vo2 =ð2gÞ=fzð1 hm =zÞ þ hm lnðhm =zÞg ðA12:5aÞ
In this equation Volm is the maximum expanded gas volume in the
vessel, Volp is the volume of pipeline upstream of the feeder tank,
Vo2 =ð2gÞ is the kinetic energy of flow, z is the head difference between
the vessel and the feeder and hm is the absolute minimum head in the
vessel. Subtracting from Volm an estimate of minimum volume during
the return upsurge will yield the required volume of refilling water.
Minimum volume during the refilling upsurge cannot be defined with
any precision. For instance, in Fig. 15.17 two refilling phases are
shown. The first more modest upsurge occurs while outflow was still
occurring from feeder tanks and with the head at the pumping station
strongly influenced by the feeder tank water level. The second and
larger upsurge takes place when the feeders are refilling and so is not
relevant as far as estimating volumes abstracted from the feeder
tanks. An estimate of peak pressure during the first upsurge could be
used to establish a minimum gas volume at this point and this would
then be subtracted from maximum expanded gas volume to obtain a
required refilling volume.
278
16
Discharge conditions
The discharge arrangement of outlets from a pressure pipeline system

can have a significant bearing upon the behaviour of pressure transi-
ents, especially during later stages of an event when flow has reversed
in parts of the network. While not necessarily influencing an initial
downsurge following pumping failure for instance, subsequent events
can be materially affected. One example concerns the ability of the
downstream limits of a pipeline to provide liquid for expelling air
from the pipeline or to refill an upstream pressure vessel. Some typical
examples of outfall configurations are given together with a description
of their influence on events elsewhere in a system. Examples of time-
dependent behaviour are provided in some cases.
16.1 Vertical bellmouth

A simple and common configuration at the downstream end of a pumping
main is shown in Fig. 16.1. Normally flow enters the receiving reservoir via
the vertical bellmouth exiting at or above top water level (TWL) in the
tank. When water level in the chamber is below the discharge elevation
of the bellmouth it cannot influence static head of the pipeline system,
but rather system head is dictated by the bellmouth elevation.
Suppose flow reverses during a transient event with TWL below the
bellmouth exit elevation. The relatively small diameter and storage
capacity of the vertical pipe extending to the bellmouth will permit
head to fall quite rapidly within the pipe. Changing water level and
piezometric level H in the filling connection will be governed by
equations (16.1) and (16.2):
As dH=dt ¼ AV and V þ g=aH ¼ Jþ ð16:1Þ
279
Piezometric level during filling
TWL M
As
BWL or LDO
As
A
q
H
Drawoff
V
Piezometric levels
during reversed flow
Fig. 16.1. Vertical bellmouth filling arrangement
where As is area of the air—water interface in the connection and A is

the rising main cross-sectional area. If water level reaches the rising
main level then:
As ¼ A=sin ðÞ ð16:2Þ
Depending upon the magnitude of a reversed velocity and its duration,
the vertical pipe may empty of water and air will start to enter the
upstream pipeline. Where the reversing flow is being used to expel sub-
stantial air pockets at upstream air valves or to refill a pressure vessel or
volumetric tank, the volumes required may be substantial. This can
lead to correspondingly large air volumes within the pipeline upstream
of the outfall and reduced piezometric level over downstream parts of
the system. A case study illustrating this aspect of behaviour may be
found in section 16.5 of this chapter. Such low pressures are undesirable
from a water quality standpoint as ingress of groundwater is a possibility.
If little or no refilling water is required then it is likely that the
duration of reversed flows will be modest and emptying of the vertical
pipe may not occur.
16.2 A tank or chamber of finite area

The case of a large reservoir effectively with surface area ¼ 1 has been
considered in Chapter 6 on boundary conditions. When surface area is
280
Variable piezometric level
During steady Gravity sewer

flow
M
Qout
While filling
or emptying
M
As
Ap
Minimum level Manhole or

M discharge chamber
V Rising main
Fig. 16.2. Chamber connection with gravity flow outlet
more modest, as where a tank is being filled or where flow is entering a

discharge chamber (Fig. 16.2), variations in level are more important.
In both of these cases the changing liquid surface level in the tank or
chamber has a direct influence upon the prevailing static head
operating. This boundary can be represented by the equation:
As dH=dt ¼ VAp Qout ð16:3Þ
where As is the surface area of the chamber, Ap is the cross-sectional
area of the inlet pipe and Qout is the outflow (if any) from the tank.
If H > sewer invert then Qout ¼ fnðHÞ and if H < sewer invert
Qout ¼ 0:0. In addition there is the quasi-invariant value reaching
the tank at the time of interest along a Cþ characteristic, thus:
V þ g=aH ¼ Jþ
Setting H=t ¼ dH=dt and averaging over the time increment t
then:
As H=t ¼ ðV þ Vo ÞAp =2 Qout
281
where Qout is an average over the time step, or

As ðH Ho Þ=t ¼ ðV þ Vo ÞAp =2 Qout
where suffix ‘o’ denotes conditions at the beginning of the time incre-
ment t. Eliminating V using the quasi-invariant equation:
As ðH Ho Þ=t ¼ ð Jþ g=aH þ Vo ÞAp =2 Qout ð16:4Þ
which is easily solved for piezometric level H. By substitution in the
quasi-invariant equation, V is found.
Under reversed flow it is also possible to drain a small chamber with
the consequence that discharge head is lowered to the inlet pipe level
and an air pocket can penetrate the upstream pipeline as with the
vertical bellmouth outlet.
16.3 Back-flow connection

The possibility of significant head drop at the downstream end of a
pipeline due to reversed flow has been highlighted in considering
discharge conditions of sections 16.1 and 16.2. In a treated water
system, the amount of reversed flow may be sufficient to draw down
piezometric level to the point where pressures in the pipeline become
too low. A back-flow connection may allow reversed flow without
unacceptable pressure drop. Figure 16.3 illustrates one arrangement.
A low-level outlet is provided, fitted with a check valve to prevent
inflow to the tank at low level. The filling connection enters the tank
above TWL, maintaining a piezometric level in the inlet pipeline
sufficient to ensure that the check valve remains closed under normal
flow conditions.
Suppose a pump is tripped at the upstream end of the pipeline caus-
ing piezometric level upstream of the tank to fall and flow to decelerate
and reverse. When head at the filling connection falls below the inlet
level, inflow will cease and if piezometric level continues to decline
until it is below water level in the tank, the check valve will open to
allow flow from the tank back into the pipeline. Hydraulic level at
the downstream end of the pipeline is then essentially controlled by
water level in the tank, thus maintaining positive pressures at this
end of the system. Water from the tank is provided to allow refilling
of any pressure vessels which may be present and also to permit evacu-
ation of possible air pockets through air valves. The volume within the
tank has to be sufficient to fulfil these objectives while maintaining
adequate storage to meet normal operational requirements.
282
Piezometric level
while filling
Filling connection
Top water level (TWL)

M
Inlet loss
Storage tank surface area = At
Loss in reversed
flow connection
Piezometric level
during reversed flow
Bottom water level (BWL) or
Lowest drawoff level (LDL)
M
Rising main area = A

Reversed flow connection
with non-return valve (NRV)
Outflow connection
or drawoff
Fig. 16.3. Low-level back-flow connection
The back-flow connection only influences hydraulic transient events

once flow has reversed and has no influence prior to this happening. A
possible downside to this type of arrangement would be if a pipe burst
were to occur in the upstream pipeline. Then there would be nothing
to prevent the discharge chamber emptying. An ideal arrangement
from a hydraulic standpoint would be to have an actuated valve
which opens as necessary to provide any refilling water to the upstream
pipeline. Once the calculated volume had passed through the back-flow
connection, this valve would automatically close.
283
16.4 Siphon breakers

Where a submerged discharge to a storage facility exists at the down-
stream end of a pressure pipeline system, it is not uncommon for flow
to pass through a siphon arrangement prior to exiting from the pipeline
system. Measures can be included in the shape of a siphon breaker to
prevent reversed flow from storage. Two examples of such arrangements
are considered.
16.4.1 Above-ground storage tanks

The first of these concerns steel water storage tanks at Al Haiyer in the
UAE (Fig. 16.4). At the downstream end of a pressure pipeline trans-
mission system, vertical inlet lines rise on the outside of each tank turn-
ing through 1808 to descend inside each cylindrical tank with a
submerged discharge. Piezometric levels in the filling connections can
be significantly influenced by the presence or otherwise of vacuum
breaking arrangements. Figure 16.5 shows possible variations in head,
with and without vacuum breakers.
Without a siphon-breaking arrangement and with the inlet pipes
initially fully primed, siphonic action will continue while tank water
level covers the outlet of the filling connection. While the siphon is
operating, the head just upstream of the filling connection will be
that of tank water level inlet loss, depending upon flow direction. If
the siphon is broken at a low tank water level then, depending on
Fig. 16.4. Steel tank with high-level filling connection
284
Piezometric level when WL < base

of filling connection (inflow)
Overflow Filing connection
TWL
M
Piezometric level when

WL at TWL
Piezometric level when

WL > base of
WL > base of filling connection
filling connection
M
Piezometric level when tank

BWL WL < base of filling connection
M (outflow or initial filling)
Outflow
Steel tank – without vacuum breaker
Piezometric level for

WL < breaker level (inflow)
Overflow Filling connection
TWL
M
Vacuum Piezometric level for

breakers WL > breaker level
WL < breaker
M
Piezometric level when tank

BWL WL < breaker (outflow or
M during initial filling)
Outflow
Steel tank – with vacuum breakers
Fig. 16.5. Hydraulic conditions with and without vacuum breakers
285
inflow rate, piezometric level could rise to the level of water at the top of
the filling connection. During reversed flow, piezometric level will clo-
sely follow that of the tank water level while the siphon is primed. If the
siphon is broken by a low tank water level, piezometric level will rise to
the top of the filling connection and then descend within the vertical
pipe outwith the tank according to the relationship:
As dH=dt ¼ AV ð16:1Þ
If vacuum breakers are included then during inflow the siphon will
be broken at a much higher level when tank water level falls below
the level of the vacuum-breaking arrangement. While the siphon is
operating upstream, head H closely follows tank water level. Once
the siphon has been broken, H will tend to stabilise to some extent.
At low rates of inflow, head may reach the water level at the top of
the filling connection. For higher flow rates, head will also depend on
air inflow rates at the vacuum breaker. On flow reversal with tank
water level above the vacuum breaker then the siphon will operate
and upstream head will follow tank water level. If water level is below
the vacuum breaker, water level will start at the top of the inlet pipe
and descend within the vertical line outside the tank.
16.4.2 Vacuum disconnecting valves

A discharge arrangement which may be found at the end of a treated
effluent pumping main is in the form of a siphon fitted with a
vacuum disconnecting valve (Fig. 16.6). In the system illustrated,
twin submerged outlets discharge to a tailbay or canal figure
(Fig. 16.7). This installation forms one part of the Al Ghouta Irrigation
Project which conveys treated effluent from the City of Damascus
for reuse as irrigation water in rural areas. The vacuum disconnection
valves are closed during forward flow and open when flow reverses.
When pumps are started, the static head which they experience is
dictated by the invert level of the highest point of the pipeline if the
system is fully primed or possibly by some lower water level depending
upon antecedent transient events. In any event, the discharge pipes will
be deprimed down to the tailbay effluent level below the high point con-
taining the vacuum disconnecting valve. On the upstream side of the
high point, effluent level may be at summit invert level or at some
lower elevation.
After pumps are started, air will gradually be purged or ‘scavenged’
from the system with the vacuum disconnecting valve remaining shut
286
Fig. 16.6. Siphon outfall with vacuum disconnecting valve
287
Fig. 16.7. Siphon discharge to downstream canal
288
Vacuum disconnecting valve
Piezometric level after

vacuum is broken
Tailwater (canal) H
Piezometric level during
effluent level steady pumping and
with siphon primed
Piezometric level while H

upstream limb is
filling or emptying
M
H Qu/s
V +ve
M
H
Q
Minimum Rising main

Common horizontal datum piezometric level
Fig. 16.8. Schematic of siphon with vacuum disconnecting valve
during forward flow. As air is removed, the siphon is finally primed with
piezometric level over the summit, falling to reach tailwater
level þ downstream losses (Fig. 16.8). Sub-atmospheric pressure will
prevail over the summit, thus minimising pumping head.
For a typical pumping system of the Al Ghouta Project, Fig. 16.9
illustrates changing effluent level predicted in the upstream leg of the
discharge siphon. Initially the siphon is fully primed with pumps run-
ning. After pump failure, flow rate decreases and eventually starts to
reverse. On flow reversal over the summit, the vacuum disconnecting
valve opens and the siphon is broken. Under reversed flow, the effluent
level starts to fall, slowly at first and then more steeply as the upstream
leg of the siphon starts to empty in order to supply the liquid necessary
for pressure vessel refilling. When the effluent level reaches the rising
main, piezometric level is largely stabilised for a time by the relatively
flat pipeline profile. When flow becomes positive once more, the
upstream leg of the siphon partially refills. When vessels are filling,
289
622
System primed
621
Valve opens
620
Elevation (mOD)
Water level in
619 discharge pipework
618 Upstream Minimum water level

pipework
emptying
Upstream
617
pipework
refilling
616
0.25 0.5 0.75 1.0 1.25

Time (min.)
Fig. 16.9. Head variation at siphon following pump trip
the siphon pipes are emptying to provide the vessels with the required
liquid and when the vessels release effluent, piezometric level in the
upstream part of the siphon is recovering.
Effluent level is again determined by the equation:
As dH=dt ¼ AV ð16:1Þ
with As ¼ Au=s =sinðu=s Þ while H soffit of rising main and As ¼
A=sinðÞ for H < soffit of rising main. The much shallower gradient
of the rising main usually has a stabilising influence upon water level
and piezometric level.
The effect of this oscillation in head at the siphon end of the system
can be seen in Fig. 16.10 which depicts predicted maximum and mini-
mum hydraulic levels along the rising mains. Initial downsurge after
pumps are tripped takes place while flow in the system is positive.
Along the upstream parts of the pumping mains the minimum down-
surge is developed while discharge head is controlled by the tailwater
level during siphoning. After flow reversal, head at the siphon end of
the system first rises as the vacuum disconnecting valve opens, and
then falls as the upstream limb of the siphon deprimes. Maximum
head towards the downstream end of the system is influenced by the
maximum level as the siphon is broken. As the vessels refill, head
at the downstream end of the system decreases and so peak head at
the pumping station is partly determined by the low downstream
head resulting in a reduced maximum pressure. Minimum head along
290
630
Vacuum
breaker valve
625
Elevation (mOD)
620
615 Pumping
station
610
Pipeline profile
0.25 0.5 0.75 1.0

Chainage (km)
Fig. 16.10. Envelope curves following pumping failure
downstream stretches of the pumping main is also influenced by

emptying of the siphon pipes.
Figure 16.11 shows envelope curves for a different element of the Al
Ghouta Irrigation Scheme. Maximum and minimum head lines have
been extrapolated downstream (dotted line . . . . . . . . .) to show how
the extremes of piezometric level are influenced by prevailing effluent
level at the discharge end of the system.
Maximum transient
630 hydraulic gradient
Vacuum breaker
625 valve
Elevation (mOD)

Pressure vessels hydraulic gradient
at pumping station
615
610
Pipeline profile
0.25 0.5 0.75 1.0

Chainage (km)
Fig. 16.11. Envelope curves illustrating effect of vacuum breaker valve
291
16.5 Air valve operation

To illustrate the possible influence which discharge conditions may
have upon the upstream pipeline, a sewage pumping system containing
air valves is considered. The sewage transmission system of Sharjah
Main Drainage Pumping Station No. 1 uses parallel DN 450 and DN
700 pumping mains to convey sewage a distance of over 6 km, with
the mains discharging to a modest chamber from which flows enter a
gravity sewer. Air valves were included at around chainage 4.4 km on
each main. While the exit level into the gravity sewer from the dis-
charge chamber was above the air valves’ elevation, the base of the
chamber where the rising mains enter the chamber was below the air
valves. Figure 16.12 shows envelope curves along one of the mains
for a combined pump start/trip analysis with most of the system up
until the air valves’ site subject to vacuum pressures after trip. Maxi-
mum pressures during start-up were based upon the assumption that
the pipelines were entirely primed. Downstream of the air valves,
minimum pressures remained positive.
On flow reversal at the discharge end of the system, the chamber
empties and head falls to the extent that hydraulic levels at the down-
stream chamber are less than at the air valves’ location. This means that
the air which has been admitted to the pipelines through the air valves
cannot entirely be purged from the system during reversed flow. Some
air will remain in the pipeline when pumps are restarted and this will
alter the start-up transient analysis.
35
30
il (mASL)
25
hmax
hmin
Elevation (mASL)
20
15
10
0
301.5
603.0
904.5
1206.0
1507.5
1809.0
2110.5
2411.9
2713.4
3014.9
3316.4
3617.9
3919.4
4220.9
4422.3
4725.2
5028.0
5330.8
5633.6
5936.4
6144.9
–5
Chainage (m)
Fig. 16.12. Envelope curves for sewage rising main with air valves
292
35
d/s vessel Start 450 Mid 450
30
25
20
Head (mASL)
15
10
0
0.276
19.596
38.916
58.236
77.556
96.876
116.196
135.516
154.836
174.156
193.477
212.797
232.117
251.437
270.757
290.077
309.397
328.717
348.037
367.357
386.677
405.997
425.317
444.637
463.957
483.278
502.598
521.918
541.238
560.558
579.878
599.198
–5
Time (s)
Fig. 16.13. Head variation for combined pump start/trip analysis
There is also the question of alleviating the severe vacuum pressure

conditions which were found to occur after pump trip, between the
pumping station and the air valves. A pressure vessel was included
at the pumping station for this purpose. With the vessel in place,
Fig. 16.13 shows head variations after two pumps were started in
sequence with 30 s between each pump being operated. After around
212 min. essentially steady flow had been achieved and the pumps
were tripped together. Pump start still produces the maximum system
pressure and air valves continue to operate as before. Head variations
at the pumping station, at the start of the DN 450 main and at the
mid-point of this main, are shown in this figure. When flow reverses,
the discharge chamber at the downstream end of the system supplies
liquid to remove air which had entered through the air valves. The
air valves in turn were used to allow water to flow to the pumping
station to recharge the pressure vessel. The end result is that the final
system head is relatively low as the discharge chamber is almost emptied
and the air valves remain open with air pockets in each pipeline.
16.6 Summary of influence of discharge arrangements

Generally speaking, head conditions at the downstream end of a system
will not significantly influence the initial rarefaction pressure wave in a
rising main after a pumping failure. At later stages, especially after flow
reversal has occurred, the configuration at the discharge end of a
293
network may start to influence events. If flow reversal at the outfall

produces a decrease in piezometric level at this point, the strength of
a compression wave travelling upstream and subsequent upsurge pres-
sures at a pumping station can be significantly affected. The refilling
time of vessels will be more prolonged and filling may be incomplete.
If air valves have operated during the initial fall in head after pump
trip, the time of air venting will be increased by any fall in downstream
head at the outfall. It is even possible for downstream head to decline to
the point at which air expulsion can no longer take place. This creates a
situation in which it is restarting of pumps which will finally purge any
remaining air from the system. Pumps may also be restarting against a
reduced static head as a consequence of the lowering of downstream
piezometric level.
Detailed consideration of discharge conditions may be especially
important when making comparisons between prediction and field
observation. Timings of events and extremes of head and flow may be
significantly affected by variations of piezometric level at the down-
stream end of a pipeline.
294
17
Air valves
Alleviation of sub-atmospheric transient pipeline pressures commonly

requires the introduction of an additional supply of fluid. From a cost
point of view, it is an attractive option to use pipeline fittings which
are already present for other purposes to achieve this result. The air
valve presents this opportunity. The usefulness of these valves in
assisting to alleviate surging action comes from their ability to admit
substantial quantities of air when piezometric level falls below the air
valve elevation. Following a pumping failure for instance, as discharge
from pumps decreases, the air inflow augments the diminishing flow
from the pumping station so that downstream of the air valve, flow
deceleration is not as great as that upstream of the valve. This reduced
deceleration implies that subsequent head changes will be correspond-
ingly lower, thus alleviating minimum pressure conditions.
These valves are commonly installed along a pipeline in an arrange-
ment intended primarily to allow easy emptying and filling of the pipeline
during maintenance and in the case of double-orifice valves (DOVs) to
facilitate venting of pressurised pockets of air. Figure 17.1 illustrates the
main components of an ‘APEX’ DOV manufactured by Glenfield Valves
Ltd. The valve has an orifice at the top which allows for large air flows
through the valve with a modest head drop. Flow ceases when the
float rises to seal against the orifice. A second much smaller orifice
with a lever attached is used to control outflow of air under pressure.
The smaller orifice is adjustable. Usually these valves are of a self-
acting pattern and so may also operate to admit air during a transient
event if piezometric level at the valve falls below the operating elevation
of the air valve. If this occurs, air will enter the pipeline, producing a
discontinuity in flow past the air valve connection. Figure 17.2 illustrates
the sequence of operation of an air valve.
295
Cowl
Cover
Orifice cover Seat ring
Seal ring
Orifice bracket Float guide
Sealing face
Float
Fulcrum pin
Body
Float and lever
Body
Adjusting
screw
Fig. 17.1. Double-orifice air valve
1 Venting 2 Filling 3 Closure
4 Sealing 5 Pressurisation 6 Relieving
Fig. 17.2. Operating sequence of sewage air valve
296
Air valves
17.1 Normal air valve locations

Air valves are a feature of the majority of water and sewage systems and
the distribution of these valves is largely dictated by the longitudinal
profile of each pipeline. Figure 17.3 illustrates locations in which it
may be desirable to include air valves.
(a) A double-orifice valve may be included downstream of a pumping
station to remove air which may be entrained at a pump’s suction
intake. Following pumping failure, an air valve at this location
HGL
DOV DOV
DOV DOV
PS
(a) (b) (c)
SOV SOV SOV
(d) SOV
SOV
SOV
(e)
SOV
SOV
SOV
(f)
800 m 800 m
DOV
M M
DOV
(g) (h)
DOV = double-orifice valve, SOV = single-orifice valve
Fig. 17.3. Typical air valve locations
297
may operate to admit a substantial volume of air depending upon

operating head conditions.
(b) At a defined summit on a pipeline, entrained air may accumulate
and if not removed will eventually lead to a reduction in efficiency
of the system. A DOV will allow air to be vented under pressure
through the small orifice. After a pumping failure, if piezometric
level should fall to the valve operating level then air inflow will
occur through the large orifice. A valve at this location is important
for emptying and filling of the adjacent sections of pipeline.
(c) The locations illustrated are similar to (b) but represent a high
stretch of main, say over a plateau, and the valves serve similar
functions.
(d) This represents long stretches of main of uniform gradient. Air can
migrate along a pipeline of shallow gradient and the ability to vent
this air is useful. Typically an interval of 800 m or so might be
allowed between neighbouring valves.
(e) As for (d).
(f ) As for (d).
(g) In a gravity system an isolating valve may be provided at the
upstream end of the main. A DOV downstream of the isolating
valve is necessary for emptying and filling of the pipeline and may
also vent any air entrained at the inlet.
(h) A DOV may usefully be included at inlet to a tank or reservoir
where a valve is provided to shut off inflow. In the event of flow
reversal with the inlet valve closed, vacuum pressures would
occur in the supply pipeline, were a means of ventilating the pipe-
line not provided.
It should be noted that some pipelines may have no air valves

installed. If, in a pumping main for instance, the pipeline rises continu-
ously to the discharge point, then any air can be expected to migrate
downstream and eventually to escape through the pipe discharge. Simi-
larly, some gravity mains are devoid of air valves by virtue of their
profile. Where flow velocities are high then it may be assumed that
all air will be carried along with the flow and escape at the downstream
end of the system.
17.2 Air valves for surge alleviation

One of the attractions of air valves as a means of surge alleviation arises
from their presence along a pipeline for operational purposes. No extra
298
Air valves
cost is incurred by their use in hydraulic transient suppression unless

modifications are required or possibly extra valves.
There are many instances where, knowingly or otherwise, large- or
dual-orifice air valves operate during low transient pressure conditions.
The inflow of air which occurs has the effect of preventing substantial
negative pressures from developing in the vicinity of the valve at least.
This may appear to be good practice but it is not without its dangers.
Up until the time when the piezometric level at an air valve con-
nection has fallen to the point at which the valve will operate then
the valve remains shut and has no influence upon the propagation of
the pressure wave, with flow and rates of deceleration upstream and
downstream of the valve connection being equal.
The rate of air inflow is given by the relationship between piezometric
level, valve operating level and flow rate. Suppliers should be able to
provide the necessary data, usually in the form of a chart which can
be digitised for use in a computer model. Figure 17.4 is an example of
the air inflow—outflow relationship for a large orifice and also for the
small orifice where this adjustable orifice is at its full-open setting.
The air flow rate is expressed in units of free dry air at standard tempera-
ture and pressure (STP). STP are at an ambient temperature and
atmospheric pressure. In computations it is necessary for the computer
model to take cognisance of the effect of compression/expansion of air
in the pipeline. Air volume within the pipeline at the prevailing pres-
sure may be calculated approximately using a polytropic relationship:
pabs Voln ¼ constant ð12:2aÞ
When an air valve commences operation (Fig. 17.5), writing the
equation balancing flows upstream and downstream of the valve
connection then:
Vu=s Au=s þ Qair þ dVolair =dt ¼ Vd=s Ad=s ð17:1Þ
where Qair is air flow rate with inflow assumed þve. Air volumes
correspond with prevailing pressure in the pipeline at the valve connec-
tion. As upstream flow rate Vu=s Au=s diminishes following a pumping
failure, the inflow of air acts to supplement this upstream flow to sustain
downstream flow rate Vd=s Ad=s . Thus the rate of deceleration down-
stream of the valve becomes smaller than that upstream. This reduction
in the rate of deceleration acts to suppress the pressure transient propa-
gating downstream of the air valve, with piezometric level at the valve
itself being maintained at a higher level than otherwise. This is due to
the substantial rate of air inflow which can be achieved through the
299
Q = flow rate (m3/s)

2.0 Dp = differential pressure across orifice (bar)
p = absolute pipeline pressure on outflow
1.8 and absolute atmospheric pressure on inflow
Pipeline gauge
pressure (bar)
1.6
1.4
Small orifice – outflow only
1.2
1.0 Supercritical flow
0.8 Subsonic flow equations
0.6
Q = 1.02 ¥ ÷(Dpp) Q = 48 ¥ ÷(Dpp)
0.4
Outflow
0.2
Inflow
Large orifice – inflow and outflow
40 30 20 10
10 20 30 40 50 60 70 80 90
Discharge of free dry air in STD m3/min
0.2
Inflow
0.4
Vacuum gauge
pressure (bar)
0.6
Supercritical flow
0.8
Fig. 17.4. Air flow characteristics of double-orifice air valve
large-orifice valve. Completion of the solution is accomplished using the

quasi-invariant relationships along the Cþ and C characteristics:
for the Cþ path:
Jþ ¼ Vu=s þ g=aH
and for the C path:
J ¼ Vd=s g=aH
The appearance of the air pocket below the valve connection acts as
an internal boundary from which strong reflection of pressure waves will
occur, much as from a reservoir, although with a variable pressure
acting on the water surface rather than a constant atmospheric pressure
as is the case with a large tank or reservoir.
Maintenance of piezometric level at the valve, at a higher level than
otherwise, will act to produce greater rates of deceleration upstream and
300
Air valves
Air valve
Air inflow Qair
Air pocket (Vol )

M
Pipe cross-section = A
Vu/s Vd/s
+x
Dt
Dx Dx
Space << Dx
Fig. 17.5. Schematic of air valve operation
a more rapid flow reversal in that section of main upstream of the valve.
In a pumping main where air valves have been provided, it is possible
that some of these valves will operate following a pumping failure. This
is particularly likely if no special surge protection has been included at
the pumping station to prevent this from happening. After the pumping
failure, as the negative pressure wave or downsurge propagates into the
rising main, the hydraulic gradient at an air valve may fall to the point
where the valve opens to allow inflow of air.
Where a number of air valves along a pipeline operates during the
downsurge following pumping failure then air pockets will be developed
in the pipeline below the connection to each operating valve.
Successive air valves will each act to further suppress the downsurge
so that the minimum piezometric line may appear as shown in
Fig. 17.6. For each valve which functions, deceleration of flow in the
section of line downstream of that valve is < deceleration rate upstream
of the valve connection. Suppression of the pressure transient
downstream of air valves has been achieved by providing alternative
301
= air valve
Successive positions
of pressure wavefront
Pipeline
profile
Zones of transient
sub-atmospheric pressure
PS
Minimum transient piezometric levels without air valves
(ignoring gas release and possible column separation)
Fig. 17.6. Illustration of pressure wave propagation with air valves
sources of fluid to supplement the reducing flow upstream of each valve.

Pressure transient alleviation has been accomplished around the air
valve and in the downstream section of pipeline. At the same time,
deceleration gradient has been increased along the upstream section
of main, above the value which would occur if the air valve were not
present.
In Fig. 17.6 an indication has been given of the possible position of
the minimum transient piezometric level along the main without air
valves being present. The influence of the air valves in alleviating the
worst sub-atmospheric pressures is evident.
17.3 Events following flow reversal

Some time after the initial downsurge, flow in the system will have come
to rest in a section of main between adjacent operating air valves. A
situation is developed where air pockets are present below each valve
which has functioned and piezometric level on each side of an air
valve will be close to the level of the air—water interface. If air
302
Air valves
Upper air valve with

large inflow capacity
Air pocket
Head drop through
upper air valve = KinV 2/(2g)
M
Hydraulic grade line (HGL)
using small orifice
at lower air valve
HGL using
large orifice at
lower air valve
V+
Head drop using small orifice DH
of lower air valve = KoutV 2/(2g)
Lower air valve

with air outflow
Pipeline diameter = D
and length = L
Head drop using large orifice

of lower air valve = KoutV 2/(2g)
Air pocket volume = Vol
Fig. 17.7. Definition sketch for simple air valve analysis
inflow/outflow has ceased at a valve then pressure will be atmospheric

within the air pocket below the valve. Residual pressure wave oscilla-
tions will usually be present in the sections of main between the air
pockets but for simplicity assume that these are negligible. The piezo-
metric line in the section of main between any pair of adjacent air
pockets will appear as shown in Fig. 17.7, with the liquid column
having just come to rest. The liquid in this part of the main is not in
equilibrium as there is an adverse hydraulic gradient between the air
pockets. Steepness of this gradient is a function of the difference in
elevation between the air pockets, head drop at inlet and outlet from
the air valves and the length of main between the valves. Magnitude
of this gradient can considerably exceed that of the steady-flow
hydraulic gradient where the pipeline is laid over undulating terrain.
Under the action of this adverse gradient, flow will develop from the
higher pocket to the lower with the upper air pocket tending to grow in
volume as air flows into the line through the upper valve while the lower
pocket reduces in volume as air is vented. With large-orifice air valves,
303
air outflow rates are substantial and the head loss through a valve will
remain relatively modest.
While the air valves may be instrumental in alleviating the initial
downsurge, in many instances the ‘secondary’ pressure transients
produced by closure of air valves can be more severe than the initial
downsurge after a pumping failure. This problem arises because of the
substantial hydraulic gradient which can develop between neighbouring
operating air valves where inflow/outflow of air is via a large orifice.
Only a modest differential head occurs across an operating air valve
while using this large orifice and so the air valve has little influence
on the development of reversed flow in the main.
Many air valves are also fitted with a second ‘small’ orifice valve
intended for venting of small accumulations of air under pressure.
Such modest quantities of air may enter the pipeline, through a pump
suction for instance, tending to accumulate at high points on a line
leading to an eventual decrease in overall efficiency of transmission.
Supposing that air outflow is only through the small orifice. Under
the initial adverse hydraulic gradient between two air pockets, flow
starts to develop towards the lower pocket and air begins to evacuate
through the small orifice of the lower air valve. As air outflow develops,
air pressure within the lower pocket increases and the volume of air
decreases due to compression as well as venting.
The rising head at the lower air pocket will cause hydraulic gradient
to flatten between the adjacent air pockets, since:
Hupper Hlower ¼ head loss at valves
þ pipeline resistance between valves
þ inertial head
Head in the pipeline at the upper air valve will not drop significantly
below atmospheric pressure if the air pocket is increasing in volume as a
result of continued air inflow. If a still higher air valve is present further
downstream, it is possible that air may also be flowing out of the middle
air valve, in which case the head at this valve will also rise significantly.
In contrast with the earlier illustration when the large orifice was used
to discharge air at the lower valve, use of the small orifice restricts rates
of outflow and flattens the hydraulic gradient thus reducing head
available to accelerate flow and to overcome pipeline resistance.
Accordingly dV=dt is smaller and the eventual flow velocity when the
valve shuts is also limited. Upsurge after valve closure is correspondingly
lower because the differential velocity is smaller.
304
Air valves
Velocities outwith the section of main under examination in Fig. 17.7

are assumed negligible. Also neglected is the deformation of air, water
and pipe wall. Rigid column theory can thus be applied.
The overall head difference between the air valves is made up of the
head drop as air flows into the line at the upper air valve; the head drop
as air exits from the pipeline through the lower air valve; minor losses
and resistance losses along the section of main; and inertial effects as
the water column accelerates towards the lower air valve. In the
absence of compression/expansion effects, volume flow through the
air valves will be of equal magnitude to pipeline flow.
The overall head difference Hupper Hlower ¼ H can then be
written:
H ¼ L=g dV=dt þ FV2 ð2:1Þ
where
n X o 2
FV2 ¼ fL=D þ KL V =ð2gÞ ð17:2Þ
P
where KL is the sum of pipeline minor loss and inlet and outlet loss
coefficients at air valves. Rearranging equation (2.1) then:
dt ¼ L=fgðH FV2 Þg dV
from conservation of volume at the lower air valve:
dVol=dt ¼ AV or dVol ¼ AV dt
where pipe cross-section area A ¼ =4D2 . Substituting for dt then:
dVol ¼ LA=ðgHÞV dV=½1 F=ðHÞV 2
p p
setting
p x ¼ fF=HgV, then V ¼ fH=Fgx and dV ¼
fH=Fg dx. The pipe volume ¼ LA ¼ Volp . Substituting then:
dVol ¼ Volp =ðgFÞx dx=ð1 x2 Þ
If z ¼ 1 x2 therefore dz ¼ 2x dx, or x dx ¼ 12 dz. Substituting
then:
dVol ¼ Volp =ð2gFÞ dz=z
Integrating
Vol ¼ Volp =ð2gFÞ lnðzÞ þ C
or
Vol ¼ Volp =ð2gFÞ lnð1 F=HV 2 Þ þ C ð17:3Þ
305
When Vol ¼ VolðmaxÞ (at time ¼ 0) then V ¼ 0 ( just as flow reverses)

so that:
VolðmaxÞ ¼ Volp =ð2gFÞ lnð1Þ þ C or C ¼ VolðmaxÞ
then
Vol ¼ VolðmaxÞ þ Volp =ð2gFÞ lnð1 F=HV2 Þ ð17:4Þ
When the lower air valve shuts, volume is reduced to zero and velo-
city has reached its maximum so that:
X
2
lnð1 F=HVðmaxÞ Þ ¼ VolðmaxÞ =Volp fL=D þ KL
or
h X i
2
1 F=HVðmaxÞ ¼ exp VolðmaxÞ =Volp fL=D þ KL
Finally,
pn X
VðmaxÞ ¼ 2gH= fL=D þ KL
h X io
1 exp VolðmaxÞ =Volp fL=D þ KL ð17:5aÞ
If VolðmaxÞ ! 1,
pn X o
VðmaxÞ ! 2gH= fL=D þ KL ð17:5bÞ
Equation (17.5b) is the result to be expected when an equilibrium

condition has been reached. Equation (17.5a) shows that velocity
developed in a section of main is dependent upon the length, gradient
and size of air pockets. Where piezometric gradient between air valves is
large and air pockets are similarly large then there may be time for
velocity to achieve a maximum value where the difference in piezo-
metric levels is entirely absorbed in overcoming pipeline resistance
and also head losses at the adjacent air valves, equation (17.5b).
When air pockets are small and gradients are lower there may be
insufficient time for equilibrium to be attained and velocities will be
smaller than that given by equation (17.5b).
For a hypothetical section of main, length ¼ 1000 m and dia-
meter ¼ 1000 mm, friction factor f ¼ 0:02 and with a 10 m overall
head difference between the upper and lower air valves, plotting
VðmaxÞ as a functionPof initial air volume VolðmaxÞ for different values
of the parameter KL =ð fL=DÞ then the curves of Fig. 17.8 were
obtained. These curves demonstrate that as air pocket volume is
306
Air valves
Maximum velocity when

the air valve shuts (m/s)
V(max) SKL/(fL/D)
3.5
0
3
2.5
1
2
1.5 3
7
1
15
0.5 31
0
0.1 0.316 1.0 3.16 10.0 31.6 100.0 316.0
Maximum air volume, Vol(max) (m3)
V(max) = ÷{2gDH/(fL/D + SKL)(1 – exp[–Vol(max)/Volp(fL/D + SKL)])}
DH = 10 m; L = 1000 m; D = 1000 mm; f = 0.02; Volp = pD 2L/4
Fig. 17.8. Maximum reversed velocity plotted against initial air volume at air
valve
increased, the time available to accelerate the reversed flow is also

increased, leading to an increased velocity when the valve finally
shuts. With a small air pocket, limited time is available for acceleration
and so final velocity remains low. As overall system resistance is
increased, the remaining head available to accelerate the reversed
flow is lower and the final velocity is also reduced. Clearly, for a
given system pipeline, resistance is fixed and so the only available
means of increasing overall system resistance is by adjusting the air
flow characteristic of the air valve. The results of this ‘rigid column’
analysis become more approximate as resistance is increased. As head
drop through the lower air valve rises with greater resistance to flow
through the orifice, the pressure within the air pocket is increased
and compression of the air mass becomes more important. Volume
flow rate through the air valve can no longer be approximated to pipe-
line flow. This aspect will be considered in Chapter 18. Figure 17.8 has
served to demonstrate that throttling of air outflow has the effect of
limiting reversed velocity. This finding has been used to produce air
valves which permit a large air inflow rate but which impose an
increased resistance during air outflow. The vented non-return attach-
ment for the Glenfield ‘APEX/EPEX’ valve employs this principle.
Other manufacturers also provide this option.
307
17.4 Air valve closure

The importance of the velocity developed while air is being vented at an
air valve is demonstrated by events subsequent to closure of the valve
on completion of air removal. During the process of air venting, a
difference in flow rate will develop in the sections of pipeline upstream
and downstream of the valve. At the instant when the air volume is
completely removed and the air valve shuts, differential flow across
the air valve connection is abruptly eliminated.
Figure 17.9(a) shows an air valve located in a section of pipeline.
Immediately before valve closure:
Vu=s þ g=au=s Ho ¼ Jþ and Vd=s g=ad=s Ho ¼ J
where Vu=s , Vd=s and Ho represent flow conditions just before valve
closure. Just after the air valve shuts:
V þ g=au=s H ¼ Jþ and V g=ad=s H ¼ J
or
H ¼ 1=gð Jþ JÞ=ð1=au=s þ 1=ad=s Þ
and
V ¼ ðau=s Jþ þ ad=s JÞ=ðau=s þ ad=s Þ ð17:6aÞ
where V and H represent velocity and total head in the pipeline at the
air valve connection. Substituting for Jþ and J then:
H ¼ 1=gðVu=s Vd=s Þ=ð1=au=s þ 1=ad=s Þ þ Ho
and
V ¼ ðVu=s au=s þ Vd=s ad=s Þ=ðau=s þ ad=s Þ ð17:6bÞ
Where acoustic velocities upstream and downstream of the valve are
the same, equations (17.6b) reduces to:
H ¼ a=ð2gÞðVu=s Vd=s Þ þ Ho and V ¼ 12 ðVu=s þ Vd=s Þ ð17:6cÞ
The abrupt upsurge produced by closure of the air valve is dependent
upon the strength of the differential velocity Vu=s Vd=s just before
valve closure. Suppose in a pipeline the acoustic wavespeed upstream
and downstream of the air valve is 1000 m/s and the differential velocity
Vu=s Vd=s ¼ 1:0 m/s. If venting is through the large orifice then
pressure head in the pipeline below the air valve will be close to
atmospheric. The head rise H Ho on valve closure will be:
H Ho ¼ 1000:0=ð2 9:81Þ 1:0 ¼ 51 m
308
Air valves
Wave fronts after

closure of air valve
V–a V+a
Ho H Ho
V
Air valve
+x
(a)
Wave front after air valve

closure at shut valve
V+a
Ho
Shut valve, e.g. NRV V=0
+x

(b)
Wave front produced on

reflection from shut valve
V+a Pressure wave Wave front on reflection

fronts produced by from air pocket, tank
air valve closure or reservoir
V–a V+a V–a

Hr
M
H
Ho Ho
V
V=0
Shut valve,
e.g. NRV +x
(c)
Fig. 17.9. Head conditions (a) and (b) just after air valve closure and (c) as
pressure waves are reflected
309
Suppose the air valve is just downstream of a shut valve, such as the
check valve at a pumping station (Fig. 17.9b). A single characteristic
will arrive at the air valve connection at the moment of closure, then:
J ¼ Vd=s g=aHo
Just after valve closure, velocity ¼ 0 and head will be H, so that:
0 g=aH ¼ J ¼ Vd=s g=aHo
giving
H ¼ Ho aVd=s =g ð17:7Þ
Considering a wavespeed of 1000 m/s and a reversed velocity
Vd=s ¼ 1:0 m/s then the head rise after valve closure will be:
H Ho ¼ 1000:0 ð1:0Þ=9:81 ¼ 102 m
This is double the amplitude of the head rise at an air valve located
along a pipeline. For a valve along the pipeline, the abrupt pressure
rise after air valve closure will propagate both upstream and downstream
from the now closed valve as compression waves (Fig. 17.9c), thus
affecting adjacent sections of main. Subsequently the pressure waves
will be reflected from closed check valves, other still-existing air pockets
and discharge points to tanks and reservoirs. The result can be a
complicated pattern of hydraulic transient behaviour in the pipeline
as a whole.
17.5 Case study of a sewage pumping station

Wave reflections of a pressure wave produced by an air valve closure
can cause that valve, and other neighbouring valves, to reopen and
close several times. Consider a relatively straightforward sewage
pumping scheme in the State of Qatar whose pipeline profile is
shown in Fig. 17.10. Pumping failure was predicted to open air valves
at the pumping station and at chainage 2.19 km along the pipeline.
Vacuum pressures were anticipated over almost the entire main with
an air pocket in the pipeline around chainage 2.19 km. Subsequent
closure of this air valve after a period of venting through the large orifice
produced peak transient piezometric levels of þ70 mHSD.
The variation of air volume contained in the pipeline at the pumping
station is shown in Fig. 17.11. The valve initially opens when head falls
after pumping failure. The subsequent period of air admission occurs as
a consequence of pressure wave reflections following closure of air
310
Air valves
Envelope curves of maximum and minimum head along the rising main
70
60 Maximum piezometric level
50
Elevation (mHSD)
40
30
20
Pipeline profile
10
–10
1 2 3 4
Chainage (km)
Fig. 17.10. Envelope curves with large-orifice air valve operating
valves when the first air mass has been expelled. Air volume variation at
chainage 2.19 km is more complicated with the air valve opening and
closing repeatedly (Fig. 17.12). Transient head variation at the
pumping station (Fig. 17.13) shows the initial fall in head after trip
followed by a long interval during which air first flows into the line
and then is removed from the pipeline, culminating in an abrupt
Air volume variation at pumping station air valve
3500
3000
2500
Air volume (l)
2000
1500
Air valve opens Air valve opens
1000
Air valve shuts
500 Air valve shuts
Air valve opens
0.5 1.0 1.5 2.0 2.5

Time (min)
Fig. 17.11. Air volume variation at PS air valve after pump failure
311
Air volume variation at chainage 2.19 km

200
150
Air valve opens Air valve opens

Air volume (l)
100
Air valve shuts Air valve shuts
50
0.5 1.0 1.5 2.0 2.5

Time (min)
Fig. 17.12. Air volume at pipeline air valve
head rise after around 1.5 min. Subsequent reflection of the pressure
wave produced by this sudden head rise causes the valve to reopen
and a second period of air inflow and outflow to occur. This second
air flow interval is concluded with a further steep head rise, at around
2.5 min, to be followed by a third interval of air inflow. At chainage
2.19 km, head variations are more complicated (Fig. 17.14). Each
Head variation at pumping station
70
60
50
Elevation (mHSD)
Steady pumping head

40
30 Pump trip
20 PS air valve open PS air valve open
10
0
PS air valve shut PS air valve shut
0.5 1.0 1.5 2.0 2.5

Time (min)
Fig. 17.13. Head at pumping station after trip
312
Air valves
Head variation at chainage 2.19 km
70 Effects of PS
air valve closure
60
50
Elevation (mHSD)
Pump trip
40
Effects of air valve closure
at chainage 2.19 km
30
20
10
0.5 1.0 1.5 2.0 2.5

Time (min)
Fig. 17.14. Head along pipeline after trip
closure of this air valve produces an abrupt head rise and compression
waves whose reflection causes a further valve reopening. Behaviour at
this location is however dominated by the pumping station air valve
closures which produce the steep rise in head after around 1.5 min
and 2.5 min.
It is the uncontrolled acceleration dV=dt in the pipeline, between the
neighbouring open air valves which is responsible for the magnitude of
the differential velocity present upstream and downstream of the air
valve just as that valve shuts. Subsequent equalisation of velocities in
the main after air valve closure produces the shock pressure rise
predicted.
In view of the beneficial effect of throttling air outflow, it may be
thought that all air valves which may operate during a transient
event should have an air outflow regulating device fitted. A short-
coming of this approach can be the time taken for a large volume of
air to be expelled. During pipeline filling operations, if air outflow is
throttled at every valve in a system, the time taken to fill the main
could become excessive.
Where a number of air valves function during the initial downsurge
after a pumping failure, the subsequent behaviour of hydraulic transi-
ents can become quite complicated. The use of air valves can be
beneficial in relieving negative pressures as a result of pump trip.
Inflow of air via the air valves reduces rates of deceleration downstream
313
of a valve. Steepness of hydraulic gradient upstream of an operating valve

will tend to increase dV=dt in the upstream section of main. When flow
reverses it is important to control rates of air outflow and the magnitude
of the differential velocities developed at air valves prior to closure. In this
way, serious secondary pressure transients can be avoided.
17.6 Pump restart with air in a pipeline

Circumstances can arise where a pipeline may still contain pockets of air
at the time when a pump is started. This can be a consequence of pipe-
line typography, or it may arise because insufficient time has been
allowed to elapse between a previous pump trip and the restart of a
pump. Restarting of a pump with air remaining in the system may
possibly result in hydraulic transient and system operation problems.
It is necessary to ensure that any study of pressure transients in circum-
stances of pumping failure is continued for a sufficient length of time to
ensure that all possible venting of air has been completed. Built into any
pump control philosophy should be an adequate time delay between
stopping and restarting to allow air outflows to be completed.
If it is necessary to start a pump with an air pocket in the pipeline, the
same danger can arise as when a reversing water column purges air
through a typical large-orifice air valve. In this case the differential
flow rate occurs because the upstream pump discharge considerably
exceeds the downstream flow, which may be at or near zero.
Consider the Sharjah Main Drainage Pumping Station No. 1 and its
associated pipelines as already introduced in Chapter 16. Two mains
run in parallel: one is DN 450 and the other DN 700. As already
discussed, the demand for water to refill the pressure vessel at the
pumping station and the drawdown in level at the discharge chamber
does not permit completion of air venting through the air valves
around chainage 4.4 km on each pipeline. Appreciable air volumes of
5 m3 and 35 m3 remained in the DN 450 and DN 700 mains respec-
tively when pumps were restarted.
With standard large-orifice air valves, air removal occurs relatively
quickly and with a relatively low pressure in the pipelines at the air
valves’ connection. This process is shown in Fig. 17.15. Air removal
from the DN 450 main has been completed before 1 min and from
the larger DN 700 main before 2 min. As regards head variations,
typical behaviour for the DN 700 main is shown in Fig. 17.16. Head
is initially varying fairly regularly under the control of the pressure
vessel. After about 1 min some irregularities appear caused by closure
314
Air valves
Sharjah PS No. 1
35
AV 450 AV 700
30
25
Air volume (m3)
20
15
10
0
0.828
7.452
14.076
20.700
27.324
33.948
40.572
47.196
53.820
60.444
67.068
73.692
80.316
86.940
93.564
100.188
106.812
113.436
120.060
126.684
133.308
139.932
146.556
153.180
159.804
166.428
173.052
179.676
186.300
192.925
–5
Time (s)
Fig. 17.15. Air removal on pump restart with large orifice in use
of the DN 450 pipeline air valve. At around 2 min the air valve on the
DN 700 main shuts, producing an abrupt head rise and further irregular
fluctuations in head.
Replacing the standard large-orifice air valves with restricted outflow
valves has the effects of prolonging air venting times and reducing the
secondary transient effects following air valve closure. Figure 17.17
Sharjah PS No. 1
40
35
30
25
Head (mASL)
20
15
10
5
strt 700 mid 700
0
0.828
7.452
14.076
20.700
27.324
33.948
40.572
47.196
53.820
60.444
67.068
73.692
80.316
86.940
93.564
100.188
106.812
113.436
120.060
126.684
133.308
139.932
146.556
153.180
159.804
166.428
173.052
179.676
186.300
192.925
Time (s)
Fig. 17.16. Head variation on pump restart with large-orifice outflow
315
Sharjah PS No. 1
35
AV 450 AV 700
30
25
Air volume (m3)
20
15
10
0
0.828
79.488
158.148
236.809
315.469
394.129
472.789
551.450
630.110
708.770
787.431
866.091
944.751
1023.411
1102.072
1180.732
1259.392
1338.052
1416.713
1495.373
1574.033
1652.694
1731.354
1810.014
1888.674
1967.335
2045.995
2124.621
2203.247
2281.872
2360.498
2439.123
–5
Time (s)
Fig. 17.17. Air removal using small orifice
shows the manner in which air volume changes in the DN 450 and DN
700 pipelines when pumps are started. A relatively rapid initial reduc-
tion in each air volume occurs as the pumped flow compresses the air
masses. Some pressure wave effects are present, resulting in an oscilla-
tion in air volume before an almost steady rate of reduction in volume is
attained at each site. The time to completely remove the air mass in the
DN 450 main is around 7 min, while expulsion of the air pocket in
the DN 700 pipeline takes almost 37 min. The corresponding head
variations at the start and at the mid-point of the DN 450 main are
shown in Fig. 17.18. The start-up transient has dissipated before the
air valves close. At around 7 min when the valve on the DN 450
main shuts, only a modest head rise was predicted. Similar head changes
occur in the DN 700 main. After 7 min only very small effects of the DN
450 pipeline air valve closure were predicted to occur in the larger main.
After 37 min or so, closure of the air valve on the larger main also
produces a quite modest head rise.
Throttled air valve closure on the DN 700 main produces a more
modest transient effect than an identical valve on the DN 450 main.
The same valve will produce a smaller head rise on closure as pipeline
diameter increases. To illustrate this effect, for simplicity assume that
resistance is essentially that of the air valve throttle and this can be
represented as Q ¼ Kout Habs , where Q is flow rate, Habs is absolute
pressure head at the valve connection and Kout is the valve throttle
316
Air valves
Sharjah PS No. 1
35
30
25
Head (mASL)
20
15
10
5
strt of 450 mid 450
0
0.828
19.872
38.916
57.960
77.004
96.048
115.092
134.136
153.180
172.224
191.269
210.313
229.357
248.401
267.445
286.489
305.533
324.577
343.621
362.665
381.709
400.753
419.797
438.841
457.885
476.930
495.974
515.018
534.062
553.106
572.150
Time (s)
Fig. 17.18. Head variation with small orifice in use
coefficient. Assuming that a steady venting rate has been achieved,

then for a given available head Habs :
VA ¼ Kout Habs or V ¼ Kout Habs =A
and head rise after valve closure, Hi , is given by:
Hi ¼ aV=g ¼ aKout Habs =ðgAÞ ð17:8Þ
2
or, Hi is proportional to 1=D .
When air is being vented through an air valve along the pipeline, a
positive pressure is developed within the air mass below the air valve.
The magnitude of this positive pressure is dependent upon the
resistance imposed by the air valve orifice and upon the air outflow
rate. In the case of a large orifice, the positive pressure will be relatively
modest, resulting in a hydraulic grade line (a) in Fig. 17.19 and a pump
operating point (a) as illustrated. The pump will deliver > design flow to
the air valve connection site while the downstream pipeline stretch
experiences only a small acceleration because of the very limited
pressure rise at the valve connection. The differential velocity in the
pipeline across the air valve connection will therefore be relatively
large, leading to a significant head rise on valve closure. When a
small orifice with appreciable air outflow resistance is employed, pres-
sure in the air mass rises, producing an hydraulic grade line (b) in
Fig. 17.19. The pump operating point (b) is closer to the duty point
317
H
(c)
(b)
HGL with system primed

(a)
(c)
(b)
(a)
Pump performance
HGL with restricted outflow

through air valve
HGL with free outflow

at air valve
M
Air valves
PS
Pipeline profile
Fig. 17.19. Hydraulic gradient when venting air
and flow rate from the pumping station to the air valve is < in case (a).
The increased pressure in the air mass produces a downstream hydraulic
gradient which is useful in accelerating the liquid column towards the
discharge point. When the air valve closes, the differential pipeline
velocity across the valve connection will be reduced considerably
from that with the large orifice. After steady flow has been achieved,
the hydraulic gradient will be as curve (c) with the corresponding
pump operating point (c).
17.7 Other considerations

Volumes of air drawn into a pipeline can range from a few litres to many
tens of cubic metres. In common with other hydraulic transient protec-
tion such as feeder tanks, an air valve at a high point adjacent to a
pumping station may be needed to provide the fluid required to allow
pressure vessel refilling at the pumping station.
17.7.1 Uncertainties in simulation

Prediction of pressure transients where a number of air valves operate
during the event can be readily conducted using available computer
318
Air valves
programs. Interactions between hydraulic transients produced by

closure of individual air valves can become quite complicated but this
is no impediment to the program. However, given the approximations
involved both in use of the gas law to represent behaviour of air
masses and overall air valve and chamber head loss characteristics,
calculations cannot be exact. This is especially the case where an
appreciable number of air valves are functioning. It may be advisable
to investigate the possibility of one or more critical air valves being
out of service. A critical valve is one which has an important influence
upon the progress of the transient. Due to the relatively low cost of air
valves, these have become a recognised solution to the task of pressure
transient alleviation. Prediction of their effects must be adequately
modelled and consideration given to any imperfections in the analysis,
especially where a large number of valves are involved.
17.7.2 Liquid being conveyed

Choice of air valves requires consideration of the liquid being conveyed.
When sewage is involved, the risk of a valve becoming blocked due to
matter being carried in the flow entering the valve has to be borne in
mind. Sewage air valves should have generous flow passages for liquid
in order to minimise possibility of malfunction. Figure 17.20 shows
Components
Large orifice
Sealing ring
Small orifice
Small orifice lever
Air valve float
Tappet
Air valve float guide
Air valve unit
Elevator
Guide sleeve
Main cover
Float chamber
Operating float
Fig. 17.20. Components of sewage air valve
319
Fig. 17.21. Air valve chamber
the large float chamber of the Glenfield ‘EPEX’ sewage air valve. Other
suppliers such as Erhard similarly provide a spacious float chamber for
their sewage valves.
17.7.3 Inspection
An adequate programme of inspection and maintenance of these valves
forms an important part of the overall package of protection for any
pipeline system. Figure 17.21 shows a sewage air valve chamber open
for inspection during transient tests.
17.7.4 Valve chamber and cover

Analysis usually assumes that the air external to the valve is at
atmospheric pressure, which itself may vary with climate and altitude.
The cover on any air valve chamber should allow sufficient air inflow
and outflow to satisfy this assumption and allow the air valve to
‘breathe’. Some modern highway covers for air valve chambers can be
so well sealed that during air venting the air in the chamber becomes
pressurised to such an extent that the top of the chamber eventually
fails, causing a hole to appear in the road surface or actually lifting
the road surface while the main is being charged, as described by
Burstall (1997). A more appropriate arrangement may be to have a
320
Air valves
vent pipe terminating above the ground surface so that a free inflow/
outflow from the chamber can take place and the air valve can function
as intended.
If air valve covers are not maintained and kept free from debris and
sand for instance, air flow can be inhibited and the performance of the
air valve may not be as modelled. Where a substantial number of air
valves are being used for the purpose of hydraulic transient suppression,
the possibility of malfunction will increase. Analyses conducted should
examine a range of possibilities to ensure that satisfactory transient
conditions are achieved with one or more valves not behaving as
intended.
17.7.5 Valve materials

Internal materials for floats, etc. can be thermoplastic or possibly stain-
less steel. The choice will depend to some extent on the duty, for
example the liquid being conveyed. Some plastics can be dissolved by
chemical effluents, resulting in pipeline contents escaping through a
damaged valve. Temperature should also feature in considerations. In
a hot climate, ambient temperature may reach or even exceed 408C.
If an air valve is exposed to this temperature and subject to high air
venting rates, heating within the valve may produce internal tempera-
tures of >808C. It has been known for plastic valve internals to soften
and become distorted under these conditions.
17.8 Buffer tanks and estimation of required volumes

It is quite common to find an air valve at the start of a rising main
downstream of a pumping station. While in many instances no special
provision is made to accommodate air admitted to the pipeline through
this air valve, in other circumstances a buffer tank may be included to
store this air volume offline, as shown in Fig. 17.22. In the first instance,
say for tendering purposes, it may be useful to estimate the required
volume of such a tank without having to resort to a detailed hydraulic
transient investigation. Determination of likely air volume admitted
directly to the rising main where no tank is provided can also be esti-
mated prior to any detailed study. The rigid column theory provides a
convenient means of establishing an initial air volume and thus tank
estimate.
Figure 17.23 illustrates circumstances following a pumping failure
when the transient piezometric level has fallen to the air valve operating
321
Air valve
Buffer tank
Air vol.
Rising main
Fig. 17.22. Schematic of buffer tank
level. Negligible head loss is assumed through the air valve during
inflow. It is also assumed that negligible deceleration of the water
column has occurred prior to air valve operation and that the friction
factor f and the elevation difference z remain essentially constant.
Equation (2.1) can be written:
L=g dV=dt þ fLV 2 =ð2gDÞ þ z ¼ 0 ð2:1Þ
or
L=g dV=dVol dVol=dt þ fL=DV 2 =ð2gÞ þ z ¼ 0
but
dVol=dt ¼ Q ¼ VA and LA ¼ Volp
therefore
Volp =gV dV=dVol þ fL=DV 2 =ð2gÞ þ z ¼ 0
Steady pumping piezometric line
hf = steady flow head loss

M
Minimum piezometric line

z
Air valve
V
M
Rising main
Buffer tank Diameter = D
Length = L
Cross-section = A
Pumping station
Fig. 17.23. Hydraulic gradient at buffer tank
322
Air valves
Writing e ¼ V 2 =ð2gÞ, then

Volp de=dVol þ fL=De þ z ¼ 0
rearranging,
de=ð fL=De þ zÞ þ dVol=Volp ¼ 0
and integrating,
ð ð
de=ð fL=De þ zÞ þ 1=Volp dVol ¼ constant
or
fL=DVol=Volp þ lnð fL=De þ zÞ ¼ constant ð17:9aÞ
Initial velocity is Vo and e ¼ eo ¼ Vo2 =ð2gÞ while Vol ¼ 0. Substituting
then,
0 þ lnð fL=Deo þ zÞ ¼ constant
When air volume has reached its maximum as flow downstream of
the air volume comes to rest then, e ¼ 0 and Vol ¼ Volmax . Then,
fL=DVolmax =Volp þ lnðzÞ ¼ lnð fL=Deo þ zÞ
rearranging,
fL=DVolmax =Volp ¼ lnfð fL=Deo þ zÞ=zg
and finally,
Volmax =Volp ¼ D=ð fLÞ lnf1 þ hf=zg ð17:9bÞ
Equation (17.9b) was presented by Livingston in 1969. Static head is
a function of volume, and as air volume increases z will increase. Where
overall static head is appreciable, any variation in z will be relatively
unimportant. On the other hand, if say a vertical buffer tank is selected
for a low head installation, the variation in z may have a significant
influence on overall static. In these circumstances it may be worthwhile
to insert extremes of z into the equation to assess the importance of
changes in z. These extremes could include values of static head
measured from the top and bottom of the buffer tank.
Equation (17.9b) can also be used to estimate air volume when no
tank is included but the air valve sits directly on top of the pipeline.
The variation in z is likely to be more modest in this situation, with
the air volume stored within the pipeline itself.
323
Pipeline profiles in low-head applications may have a number of high

points along the route. Should air valves operate at intermediate
summits, the effect will be to reduce the actual air volume admitted
to the pipeline or buffer tank at the start of the main, from the value
calculated from equation (17.9b).
324
18
Air and gas
In Chapter 17 it was described how the introduction of air into a

pipeline system may be used to beneficial effect, namely the avoidance
or limitation of serious vacuum during a pressure transient event. It
was also described how if not carefully regulated, serious secondary
hydraulic transients could occur following venting of air and/or gas
through air valves. The present chapter presents several case studies
to illustrate circumstances in which air or gas masses can prove
problematic.
18.1 Pump start-up with an air-filled riser

The first of these studies concerns the common example of a pumping
station in which an air column exists between the pump and a down-
stream check valve. Pumping station arrangements may loosely be
described as being of two types. First there is the wet well/dry well
configuration. In this the check valve is a short distance downstream
of the pump. The pipeline within the pumping station and along the
initial parts of a rising main usually remains liquid-filled even under
modest static head conditions. Since no air or gas should be present
in the riser column downstream of the check valve, no particular
problem of air removal is anticipated during pump start-up.
The alternative arrangement, and one which is relatively common
nowadays, is to have a wet well containing submersible pumps
(Fig. 18.1). A vertical riser extends above sump water level to a horizon-
tally placed check valve. An air valve is usually sited just upstream of
the check valve. Downstream of this check valve the pumping station
pipework may include other pump branch connections where a multi-
pump installation is involved. Figure 18.2 shows the downstream
325
Dual-orifice EPEX air valve

DN 350 swing check valve
DN 350 knife gate valve

12.5 mm tapping for
pressure transducer
350/500 tee
c.l. = 6.803 mAOD
+6.0 mAOD finished floor level

DN 350 riser
Start levels:
pump 3 = –16.025
pump 2 = –16.125
pump 1 = –16.225
Stop level pump 1,

2 and 3 = –16.99
WPL SWALLOWGLIDE
SRVS 200DS (4 off)
–17.74 mAOD
Fig. 18.1. Submersible pump arrangement
arrangement for a fairly typical submersible pump installation. In this

figure, branch pipes from two of four foul sewage pumps can be seen.
A sewage air valve is included upstream of a swing-check valve in
each of these discharge branches.
When a pump is idle, the riser upstream of the check valve will be
filled with air down to the water level in the wet well or sump. This
configuration also occurs in boreholes where the check valve is located
at the wellhead. The problem of an air-filled riser should not arise in
326
Air and gas
Fig. 18.2. Air valves and non-return valves at submersible pumping station
these borehole pumping installations where the check valve is sited

directly on top of the borehole pump.
Hydraulic transients associated with priming of an air-filled riser
pipe following start of a borehole pump was considered of sufficient
importance to warrant particular attention and earlier studies have
examined potential problems.
Downstream of each check valve, flow travels through DN 350 lines
which rise vertically to join a DN 500 pipe at elevation þ9.9 mAOD as
shown in Fig. 18.3. A single-orifice air release valve is included at the
downstream end of this gradually rising pipe. After the DN 500 pipe
descends vertically, two further foul pump connections are made
prior to start of the rising main.
A typical pump curve for the submersible sewage pumps is as shown
in Fig. 18.4 with the duty head reflecting the design static lift and also
system resistance losses. When a submersible pump or a borehole pump
is operated with an air-filled riser, the static head experienced at the
pump at start is initially zero. If a direct pump start is assumed, flow
rate at this stage can be considerably in excess of the design discharge
and the pump motor has to be able to accommodate the higher power
demand associated with this phase of start-up. The riser pipe will
327
Single-orifice air
release valve
DN 500 cross-over pipe
Rise
Odour pipe to
wet well
c.l. = +9.9 mAOD
350/500 tee for

foul pump connection
+6.0 mAOD
To rising main
DN 350 line from
foul pump
Fig. 18.3. Pipework downstream of non-return valve
H
N = 100%
Duty point
N = 75%
Loss through
air valve
Direct start
riser filling
Loss in riser
N = 50%
Z
N = 25%
Speed necessary Q
to just fill riser Duty flow
Flow when air valve closes
Maximum flow – riser empty
Fig. 18.4. Pump performance against speed relationships
328
Air and gas
progressively become filled, with liquid level and static head at the
pump increasing and discharge reducing. The air column above the
water surface in the riser will become smaller, partly through compres-
sion but mainly as a result of venting through the air valve located at
the top of the riser. Preferred location for the air valve is downstream
of the 908 bend. During the venting process, pressure rise at the air
valve should not be too great. An excessive pressure rise upstream of
the check valve may overcome the downstream static head, allowing
the check valve to crack open and air to escape into the downstream
line. Often a typical large-orifice air valve would be installed to
ensure adequate air venting capacity.
When all air or gas has finally been purged, the air valve will shut and
the differential flow rate upstream and downstream of the check valve
will abruptly be eliminated. Downstream of the initially closed check
valve, velocity is zero and the upstream velocity will be given by the
operating point on the pump H—Q curve at the moment of valve
closure (Fig. 18.4). This upstream velocity will still be greater than
design velocity. An indication of the sudden head rise associated with
the process of velocity equalisation after air valve closure can be
obtained as follows.
From the Cþ characteristic arriving upstream of the check valve
(Fig. 18.5):
Vu=s þ g=aHu=s ¼ Jþ
where Vu=s and Hu=s are given essentially by the pump operating point
(Fig. 18.4), ignoring riser losses and head loss through the air valve.
From the C characteristic arriving on the downstream side of the
check valve (Fig. 18.5):
0 g=aHd=s ¼ J
where Hd=s is the prevailing head downstream of the check valve. If no
other pump is already operating, Hd=s will be the static head. Where
pumps are already running, Hd=s will be given by the appropriate
point on the system curve.
When flow is abruptly equalised, common velocity V is given by:
V ¼ ð Jþ þ JÞ=2 ¼ ðVu=s þ g=aðHu=s Hd=s ÞÞ=2
The transient pressure H when upstream and downstream water
columns meet is given by:
H ¼ ð Jþ JÞ=ð2g=aÞ ¼ ½Vu=s þ g=aðHu=s þ Hd=s Þ=ð2g=aÞ
329
Hd/s
Air valve
Hu/s
Discharge
NRV header
Riser
Vu/s Vd/s = 0.0
Dx + Dx
C+ C–
Dt
J+ = Vu/s + g/aHu/s J– = 0.0 – g/aHd/s
From pump
Fig. 18.5. Flow conditions during air venting
or
H ¼ aVu=s =ð2gÞ þ ðHu=s þ Hd=s Þ=2 ð18:1Þ
This head rise H can be very considerable and unacceptable.
Figure 18.6 shows an example of predicted velocity in the filling riser
of Fig. 18.1. The pump reaches design speed in 1.0 s and a maximum
velocity of 3.2 m/s is developed. As the riser fills, head at the pump
delivery increases and pump operation gradually moves back up its
performance curve (Fig. 18.4), to produce a reduced flow rate. Velocity
continues to decrease gradually until the riser is filled and the air valve
shuts. The velocities upstream and downstream of the air valve con-
nection are abruptly equalised and a high-frequency transient occurs.
Eventually a steady velocity of <1.0 m/s is produced. Head changes
at the air valve connection for this 1 s pump start are contained in
Fig. 18.7. Only small changes occur until the air valve shuts when
upstream and downstream velocities are equalised. An abrupt head
rise develops reaching a peak of 180 mAOD. The high-frequency
transient dissipates quite rapidly to leave a steady operating head of
330
Air and gas
AYR solo pump op/1 s start

3.5
Delivery
3.0 d/s NRV
2.5
2.0
Velocity (m/s)
1.5
1.0
0.5
0.0
0.003
0.382
0.760
1.139
1.517
1.896
2.274
2.653
3.031
3.410
3.788
4.167
4.545
4.924
5.302
5.681
6.059
6.437
6.816
7.194
7.573
7.951
8.330
8.708
9.087
9.465
9.844
10.223
10.601
10.980
11.358
–0.5
Time (s)
Fig. 18.6. Velocity during/following riser filling
about 38 mAOD. The peak head of 180 mAOD influences the

pump delivery branch as far downstream as the 350/500 connection
(Fig. 18.3) and the head variations at the pump delivery, the air
valve connection and downstream of the NRV are shown in Fig. 18.8
for a time interval of 0.8 s around the time of air valve closure. Further
downstream in larger pipe diameters at the elevated air valve and at the
start of the rising main, velocity changes are smaller.
200
180 AV connection
160
140
120
Head (mAOD)
100
80
60
40
20
0
0.003
0.422
0.840
1.258
1.677
2.095
2.513
2.932
3.350
3.768
4.187
4.605
5.023
5.441
5.860
6.278
6.696
7.115
7.533
7.951
8.370
8.788
9.206
9.625
10.043
10.462
10.880
11.298
11.717
12.135
12.554
–20
Time (s)
Fig. 18.7. Head at the air valve after pump start
331

200
Delivery
180
AV connection
160 d/s NRV
140
120
Head (mAOD)
100
80
60
40
20
0
8.297
8.326
8.356
8.386
8.416
8.446
8.476
8.506
8.536
8.566
8.595
8.625
8.655
8.685
8.715
8.745
8.775
8.805
8.835
8.864
8.894
8.924
8.954
8.984
9.014
9.044
9.074
9.103
–20
Time (s)
Fig. 18.8. Head in the riser during/following riser filling (detail)
As in all aspects of pressure transient control where air valves are

involved, the remedy is often to control the differential velocity
which exists at the air valve connection just as the valve is about to
close. This requires some regulation of air outflow rate through the
valve or alternatively avoiding the abrupt union of water columns.
Possible means of avoiding an unacceptable shock rise in pressure
include the following.
18.1.1 A more restricted air outflow device

Using an air valve having more limited air outflow capacity will produce
a pressure rise in the air-filled riser, leading to compression of the air
mass. The pump operating at design speed will experience a greater
delivery pressure, causing it to operate at higher head and lower flow
rate than with the earlier large-orifice valve. With the lower flow rate
and velocity Vu=s , the shock pressure rise H following air valve closure
and given by equation (18.1), will also be lower, although Hu=s will be
increased to some extent.
There is a limitation on the extent to which the air valve outflow may
be throttled. If the venting rate is restricted to a large extent, head in
the air pocket below the valve may exceed the piezometric level down-
stream of the check valve, causing the valve to open slightly and air to
escape into the downstream system. Depending upon the configuration
of the check valve, some of this air may collect in the upper part of the
332
Air and gas
valve body downstream of the door. When the valve is later opened
fully after the air valve shuts, the door will travel into a region partially
filled with air. Damping of the door motion can be reduced by this air
pocket, allowing the door to open more rapidly and causing it to
violently strike its stop, possibly resulting in damage.
18.1.2 Soft-start of the pump

The pump may be operated in such a manner that it does not attain its
maximum speed until after all air has been vented through the large-
orifice air valve. Pump speed may be gradually ‘ramped’ so that from
moment to moment the H—Q curve alters (Fig. 18.4). Pump speed
remains at a rate sufficiently low that excessive velocity Vu=s does not
develop. Where Hs is the ‘shut valve’ head at the pump at design
speed Ns , if the elevation difference between the liquid surface in the
wet well and the air valve is Z then a pump speed N given by:
p
Hs =N2s ¼ Z=N2 or N ¼ Ns ðZ=Hs Þ ð18:2Þ
will be just sufficient to fill the riser and hold the liquid column in place.
A marginally higher speed will ensure a quiet air valve closure but
without opening the check valve.
Computations can be undertaken to establish the relationship
between ‘ramp slope’ dN=dt and pressure rise on air valve closure. In
this case the pump motor is capable of a maximum ramp time of
4 min and Fig. 18.9 shows predicted peak transient head at the pump
delivery, at the riser air valve connection, at the elevated air valve
and at the start of the rising main as functions of the time for the
pump to reach its design speed. A very short ramp time will not be
beneficial as it will allow the pump to achieve its maximum speed
before air valve closure and thus velocity Vu=s will also be at or close
to the maximum for the low operating head. In this case, ramp times
of 18 s or less are therefore of no benefit.
By extending the time taken for a pump to achieve design speed
till a point beyond the air valve closure time, then the velocities in
the riser can be limited and the corresponding shock pressure rise
when the valve shuts is reduced. In this instance a ramp time of
around 1 min would reduce maximum head to around one-third of
the maximum.
A two-speed motor could be used with the lower speed maintained
until riser filling has been completed without undue pressure rise.
Thereafter the pump speed can be increased to the higher speed.
333
AYR solo pump op/ramp rate results

180
Delivery AV connection
160
Elevation AV Rising main
140
Maximum head (mAOD)
120
100
80
60
40
20
0
0
9
18
27
36
45
54
63
72
81
90
99
108
117
126
135
144
153
162
171
180
189
198
207
216
225
234
Time to reach design speed (s)
Fig. 18.9. Maximum head during riser filling plotted against ramp slope
18.1.3 Use of an accumulator

An alternative surge suppression device which may be found on
borehole riser pipes is a bladder accumulator of the type illustrated in
Fig. 13.2b. The pump is started without any restriction on speed rise
and a large-orifice air valve may be included just upstream of the
check valve on the well head. The shock pressure rise, developed
when the air valve closes, travels as a compression wave to the accumu-
lator from which it is reflected as a rarefaction wave and producing rapid
pressure relief. A compression wave also travels downstream from the
air valve, causing a pressure rise which is very short-lived as it is quickly
followed by the rarefaction wave from the accumulator. Prevention of a
shock pressure increase is considered to be better than curing such a
pressure rise after it has developed. The accumulator offers a remedy
if unacceptable pressures are found during commissioning of a scheme
when other remedies are no longer viable.
18.2 Pump start with slow valve closure

Where a direct pump start is used then a further option to restrict shock
pressure rise may be to use a slow-closing air-venting valve. Consider
the situation illustrated in Fig. 18.10. A two-stage borehole pump,
Weir Pumps Ltd type SBWM 600, is intended to lift water abstracted
334
Modulo 1
Cruce
ro Cla
rines 3
ue
anq
Est rines
Cla
360
Clarines
4+
2
875
4+
2
.
V.C cion
" P. enta lara
∆10
re
40 m ta C
Alimrines = 28 510
na
Cla ro L Sannta dento
1+
U
"Ace Pla tame
io
∆10 0 Tra 476
1
R
6+
10m 1
L = 15 se
ero bal
" Ac Em Clara
∆12 ta
1
6 San
2
3
4
5
0
Tom
∆20" Aduccion
El Yai
Fig. 18.10. El Yai water transmission system

Air and gas
335
El Yai/duty pump start/velocities

5
Ch. –0 km Riser
4.5
3.5
3
Velocity (m/s)
2.5
1.5
0.5
0
0.003
0.325
0.647
0.968
1.290
1.611
1.933
2.255
2.576
2.898
3.220
3.541
3.863
4.184
4.506
4.828
5.149
5.471
5.793
6.114
6.436
6.758
7.079
7.401
7.723
8.044
8.366
8.688
9.009
9.331
9.652
–0.5
Time (s)
Fig. 18.11. Velocity at pumping station after pump start
from Rio Unare and to transfer this water at a rate of 350 litres/s either
to Santa Clara storage reservoir or directly to the nearby Santa Clara
treatment works. Pressure rating of the existing 20 in (508 mm) steel
rising main is 16 bar(g) and the pump duty head is 110 m.
18.2.1 Air venting through a standard air valve

Direct start of the pump produces a peak velocity of 4.58 m/s in the
initially air-filled riser (Fig. 18.11). Air is vented at a high rate through
an air valve sited in the horizontal section of pipeline downstream of the
908 bend at the top of the riser (Fig. 18.12). Closure of the air valve
on completion of venting produces a shock pressure rise beyond the
allowable maximum transient pipeline pressure (Fig. 18.13).
18.2.2 A butterfly valve for air venting

Installing a small butterfly valve in place of the air valve allows a quieter
start-up to be achieved. When the pump is operated, air is expelled via
the open butterfly valve connected to a backwash pipe leading back to
the suction well (Fig. 18.14). When all air has been removed, water
starts to flow through the butterfly valve back to the wet well. Due
to the greatly increased flow resistance, as liquid rather than air now
flows through the butterfly valve and backwash line, and there is a
336
Air and gas
10.670 m
3 tonne manual hoist
8.510 m
400 on 400 on TFA
6V AV
400 on
TPI RCV
SW
6.995 m
6.760 m
6.470 m
770 SD
floor opening
min W.L.
0.000 m
Inlet pipe
(by others)
–2.030 m
–3.164 m Strainer
–3.430 m
Fig. 18.12. Elevation through pumping station
El Yai/duty pump start

300
Ch. 0.0 km Riser
250
200
Head (mAD)
150
100
50
0
2.5
2.540
2.580
2.619
2.659
2.699
2.739
2.779
2.818
2.858
2.898
2.938
2.978
3.017
3.057
3.097
3.137
3.176
3.216
3.256
3.296
3.336
3.375
3.415
3.455
3.495
3.535
3.574
3.614
–50
Time (s)
Fig. 18.13. Head at pumping station after pump start (detail)
337
n
ai
m
y
er
liv
de
N
D
0
50
g
tin
is
Ex
TP1 TP2
400 DN
300 DN BF Existing
GV 300 DN 300 DN platform
TFA GV
4400 Approx.
Back flushing pipework
300 DN
New access
walkway
750
Pumpset Pumpset
1900
No. 1 No. 2
700 1300 1300 700
Fig. 18.14. Plan showing pipework and valves
shock rise in pressure at the butterfly valve connection (Fig. 18.15).

Sizing the butterfly valve appropriately allows the peak transient
pressure at this connection to be limited. Once this initial upsurge
has dissipated, the butterfly valve can be closed gradually. The shock
338
Air and gas
El Yai/duty start
250
Ch. 0.0 km Riser AV
200
150
Head (mAD)
100
50
0
2.5
2.669
2.838
3.007
3.176
3.346
3.515
3.684
3.853
4.022
4.191
4.360
4.529
4.698
4.868
5.037
5.206
5.375
5.544
5.713
5.882
6.051
6.220
6.390
6.559
6.728
6.897
7.066
7.235
7.404
7.573
–50
Time (s)
Fig. 18.15. Head after pump start (detail)
pressure rise forces the non-return valve open and flow is established in
the downstream rising main.
Following a pump trip, as piezometric level falls, the butterfly valve
automatically reopens allowing air to re-enter the riser.
18.3 Air venting through a ‘sparg’ line

Another means whereby an air-filled riser may be primed after pump
start and without the development of an excessive shock pressure rise
as the riser becomes filled is through the introduction of a ‘sparg’
line. This may take the form of a pipeline connected at the same
position as an air valve, leading back to the wet well. As the riser fills
with liquid, air passes through the sparg line to be released into the
wet well. After filling, the line then returns a relatively small proportion
of the liquid flow to the wet well. The sparg line has to be of sufficient
size that it reduces inertial pressure to an acceptable extent without
having an excessive effect on the efficiency of pumping. In the case
of sewage pumping, the sparg line may be at risk of blocking.
18.4 Gas evolution

If no air valve is placed before the check valve in a sewage system or if
an air valve becomes blocked for example, problems may still arise.
339
While the liquid column is static, gas may evolve from the column of
sewage upstream of the check valve both under septic conditions and
also because the column is under vacuum pressure. When the pump
is started and flow commences in the riser, the gas volume is
compressed. As pressure within this volume rises to match the piezo-
metric head downstream of the check valve, gas will be pushed into
the downstream line through this valve. When all gas has been removed
from upstream of the non-return valve then the column of liquid will
impact against the valve door, throwing it open violently into the
downstream fluid which is a mixture of sewage and gas. Damage to
valve doors and stops has occurred as a consequence.
18.5 Gas pockets in a pipeline

A risk may exist in pipelines generally when a pocket of air or gas is
present under transient conditions. This potential problem was
indicated by Martin (1976) with regard to the expulsion of air through
a valve while filling a pipeline using a pump. In a recently released
design manual edited by Escarameia (2005), studies of air pocket
influence on hydraulic transients indicated a possible amplification of
maximum pressures. Any amplification is strongly dependent upon
the configuration of the particular pipeline being studied but ampli-
fication was predicted by a factor of up to 260% over the equivalent
pressures without the presence of air pockets. Field observations of
Danish uPVC pipelines also showed amplification. The dependency
of peak pressure on air pocket volume was also established.
The basic problem can easily be illustrated by considering transient
behaviour in a simple pipeline containing a trapped mass of air or gas
at its upstream limit (Fig. 18.16). Supposing that the liquid-filled
sloping pipeline has previously been subject to an hydraulic transient
event, say a pumping failure, which has produced a low pressure say
8.0 mWG at the upstream end of the line. Suppose that a volume
of gas has evolved from solution during the low pressure produced by
this initial transient event. For simplicity the gas was not reabsorbed
into the liquid by the influence of rising pressure.
Under action of the adverse transient, hydraulic gradient flow will
develop from the downstream reservoir towards the gas pocket. The
strength of this reversed flow and its duration will depend upon the
ability of the system to accommodate this flow.
With no gas volume present, the only storage available within the pipe-
line is that resulting from extension of the pipe wall and from compression
340
Air and gas
10 m
DN 300
Gas pocket
Hydraulic gradient at
time of flow reversal
8m
Riser
1000 m
M
Pump and NRV
Fig. 18.16. Simple system with air pocket
of the liquid component. The amount of reversed flow and its duration are
modest in these circumstances. A short duration of reversed flow allows
only a modest time for flow acceleration to occur, resulting in a relatively
small reversed velocity. Inertial head rise is a function of this maximum
reversed velocity and so the peak pressure will also be relatively modest.
When a volume of gas is present, an additional storage component
exists, namely the volume available as the gas mass is compressed by
the increasing pressure produced by the reversed flow. This component
can be written by differentiating equation (12.2) as:
dVol=dpabs ¼ 1=nVol=pabs ð18:3Þ
Allowing reversed flow to develop, the gas volume is compressed with
rising pressure. The hydraulic gradient flattens and the acceleration
towards the gas pocket reduces and finally ceases as a positive gradient
is established. The peak pressure within the gas pocket will depend on
the maximum reversed velocity and the initial volume of gas present.
The duration of reversed flow and the maximum reversed velocity
will increase as the initial volume of gas is increased. Deceleration of
this reversed velocity will occur when the combined effects of rising
pressure in the gas pocket plus developing pipeline resistance produce
a positive hydraulic gradient within the pipeline. Even a relatively
341
160
n = 1.4
140 n = 1.2
n = 1.0
Maximum pressure head (mWG)
120
100
80
60
40
20
0
0 2 4 6 8 10 12
Initial air/gas volume (m3)
Fig. 18.17. Maximum head plotted against initial gas volume
small gas pocket allows maximum reversed velocity to significantly

exceed the corresponding velocity when no gas pocket is present.
Pressure can rise quite quickly and the eventual peak pressure is greater
than that without an initial gas volume present. As the trapped gas
volume is changed from one computer simulation to the next, the
maximum pressure following flow reversal also varies. Maximum pressure
increases substantially with initial gas charge until 450 litres and there-
after diminishes gradually. For the same sample pipeline, Fig. 18.17
depicts the maximum predicted pressure as a function of initial gas
volume for polytropic coefficient values n ¼ 1:0, 1.2 and 1.4. With
each n value, a maximum pressure occurs and as gas volume is increased
beyond this peak the maximum pressure decreases gradually. The value
of n ¼ 1:0 produced the largest maximum pressure. Larger volumes
represent the region of pressure vessel solutions where the gas charge,
while allowing larger reversed velocities to occur, also provides a
substantial cushion so that a more gradual deceleration of flow can
occur thus reducing the maximum pressure. Pipeline resistance also
plays an increasing role as reversed velocity increases.
It might be expected that as air volume is gradually reduced, the
peak upsurge pressure would increase smoothly towards a peak value
corresponding to the situation with no gas volume present. Instead it
will be observed that peak upsurge pressure occurs with a finite and
342
Air and gas
relatively small initial gas volume and that this peak can be considerably
in excess of the maximum with no gas present.
The maximum amplification of 560% shown in Fig. 18.17 over the case
when no air pocket is present is greater by a factor of about 2 than the
260% amplification reported by Escarameia (2005) for an air pocket
located partway along a pipeline. Explanation for this difference may lie
in the position of the air pocket. The compression wave produced by
amplification of pressure at a point within the pipeline, will propagate
along the main. On encountering a closed valve for example, then
amplitude of the compression wave will be doubled against the valve,
resulting in an overall amplification of 520%. The 560% amplification
was for the case of an air pocket being compressed at a closed valve
while the 260% amplification was at a pocket along the pipeline.
18.6 Throttled outflow air valves

The use of throttled outflow air valves has been presented as a cost-
effective means of alleviating vacuum pressures. Chapter 17 described
use of a throttled outflow air valve to control upsurge pressures on
completion of subsequent air venting. Pressures occurring below a
throttled air valve, when a modest air pocket is compressed by reversed
flow, can be considered. It may be possible that a similar phenomenon of
pressure amplification could occur with a pocket of air beneath the air
release valve.
Computer simulations were carried out using the same hypothetical
pipeline illustrated in Fig. 18.16 but with an air valve sited at the start of
the pipeline. The initial head in the pipeline was set at atmospheric
pressure, 0.0 mWG. Flow was just reversing after an earlier downsurge
which has caused air to enter the pipeline. The effects of a range of
initial air volumes at the time of flow reversal were examined. Outflow
capacity of the throttled air valve was initially set to that of the small
orifice of the Glenfield APEX valve and the polytropic coefficient n
was set to 1.0. Under reversed flow the air mass will reduce in
volume both through compression and also by being vented to atmos-
phere. Small initial air volumes may be completely expelled during
the initial period of reversed flow while larger volumes are not fully
removed until subsequent cycles of oscillation. Pressure rise on air
valve closure was found to be relatively modest as shown in Fig. 18.18
with peak pressures mainly influenced by compression of the air mass.
If air outflow capacity is increased, for example to ten times the
APEX capacity, little air compression occurs and all air is removed
343
Full open orifice

30
Vol = 10 litres
25 Vol = 40 litres
Vol = 80 litres
Maximum pressure head (mWG)
20
15
10
0
0.004
0.487
0.970
1.453
1.936
2.420
2.903
3.386
3.869
4.352
4.835
5.318
5.801
6.284
6.767
7.250
7.733
8.216
8.700
9.183
9.666
10.149
10.632
11.115
11.598
12.081
12.564
13.047
13.530
14.013
14.496
14.980
–5
Time (s)
Fig. 18.18. Head plotted against time for small orifice valve
during the initial reversed flow interval. Figure 18.19 illustrates that the
pressure rise on valve closure is a more significant component of the
total pressure rise.
For the system under consideration these curves demonstrate that for
a given flow capacity, peak pressure is at a maximum for a particular
n = 1.0
50
Vol = 10 litres Vol = 80 litres
40 Vol = 100 litres Vol = 150 litres
30
Pressure head (mWG)
20
10
0
0.003
0.297
0.591
0.885
1.179
1.473
1.767
2.061
2.355
2.649
2.943
3.237
3.531
3.825
4.119
4.413
4.707
5.001
5.295
5.589
5.883
6.177
6.471
6.765
7.059
7.353
7.647
7.941
8.235
8.529
8.823
–10
–20
Time (s)
Fig. 18.19. Head plotted against time for larger outflow orifice
344
Air and gas
initial air volume and that the peak pressure increases as air flow
capacity rises.
18.7 Case study of a sewage rising main

In studying a pumping system a series of pressure transient modelling
exercises was carried out varying the amount of gas evolved under
transient low-pressure conditions. It was found that there is a band of
‘critical’ gas volume which produces a peak pressure noticeably in
excess of the pressures which are predicted when the gas mass is
assumed to be either smaller or greater than this critical range of
volumes. When transient pressures are allowed to range below atmos-
pheric pressure to the point at which gas can be released from sewage
it is recommended that caution be exercised in identifying the predicted
peak pressure. Possibly extremes could be identified by exploring the
influence of varying gas volume.
Potentially the existence of a relatively small pocket of air or gas
within a pipeline may produce an unexpectedly high peak transient
pressure. As an example, consider the case of Brown’s Point pumping
station and twin rising mains. The wet well/dry well arrangement
contains vertical risers with check valves located just downstream of
the pumps at the bottom of the risers (Figs 18.20 to 18.22). The
pumping mains rise continuously to the discharge point, passing
beneath Cullercoats metro station. Since the mains rise continuously,
no air valves were included as any air or gas would be expected to
migrate downstream to the outfall over time.
Pump start-up can be achieved without noticeable adverse effect
(Fig. 18.23).
After a pumping failure, piezometric level was predicted to fall below
the pipeline elevation at the pumping station and throughout the entire
system as shown in the envelope curves of Fig. 18.24. During the period
of sub-atmospheric pressure, evolution of gas from the sewage was
audible in the upper levels of the pumping station. After flow reversal,
loud noises and vibration occurred emanating from the pipework within
the pumping station. Recorded pressure versus time records at the
pumping station (Fig. 18.25) sometimes showed pressure peaks
around double the value of those which would normally have been
expected following pressure transient analysis. These peaks were attrib-
uted to the development of a gas pocket in the pipeline at the pumping
station during the initial downsurge after pump trip. Comparisons were
made between pressure recordings and predictions for a single main in
345
1.5 Tonnes crane (existing)
Control panel
Washwater
booster set
8.00
7.10
Air relief
(existing)
0.34
TP. 2
–3.20
TP. 1
TP. 1
2500 N T.F.A.
300/2500 N T.F.A.
Part Section A–A
Fig. 18.20. Wet well/dry well pumping station elevation
use and also with both mains in service. In each case two duty pumps
were tripped together. Figures 18.26 and 18.27 show these comparisons.
The precise volume of gas evolved is not possible to predict with any
accuracy nor is the rate of gas reabsorption with increasing pressure.
346
Air and gas
Fig. 18.21. Pumps, check valves and riser
These parameters may vary to some extent from one test to another. A
modest increase in gas release rate is sufficient to produce a peak head
at the pumping station >100 m (Fig. 18.28).
The development of high pressures in the presence of a gas pocket is
essentially a matter of the time available to accelerate flow towards the
gas pocket. If the pocket is small then following flow reversal it does not
Fig. 18.22. Pipework at top of riser
347
60
Observation d/s of pump
50
Prediction d/s of pump

40
Head (mAOD)
30
20
Prediction halfway along rising mains
10
0
0.008
0.815
1.621
2.428
3.234
4.040
4.847
5.653
6.460
7.266
8.072
8.879
9.685
10.492
11.298
12.104
12.911
13.717
14.524
15.330
16.136
16.943
17.749
18.556
19.362
20.168
20.975
21.781
22.588
23.394
24.200
25.007
–10
Time (s)
Fig. 18.23. Head variations on pump start
take long for the reversed flow to compress the gas and increase pressure
within the gas volume. As pressure increases, the hydraulic gradient
causing flow to accelerate towards the pocket is first flattened and
then starts to decelerate the reversed flow. If the time available to
accelerate flow is small then the maximum reversed velocity will also
be more limited and the peak upsurge pressure will tend to remain
more modest. As gas volume is increased, the rate of pressure rise in
Brown’s Point/two pump trip/solo main/0.0008 ¥ 0.2 ¥ 0.2

90
i.l. mAOD
80
Hmax-m
70 Hmin-m
60
Elevation (mAOD)
50
40
30
20
10
0
0 100 200 300 400 500 600 700 800 900
–10
Chainage (m)
Fig. 18.24. Envelope curves on pump failure
348
Head (mAOD) Pressure (bar g)
0
1
2
3
4
5
6
7
8
9
10
0
10
20
30
40
50
60
70
80
90
0.027 07/22/99 14:29:52
07/22/99 14:31:48
0.729
07/22/99 14:33:45
1.431
07/22/99 14:35:41
2.133
07/22/99 14:37:37
2.835 07/22/99 14:47:41
3.537 07/22/99 14:49:38
4.239 07/22/99 14:51:34
4.941 07/22/99 14:53:30
5.643 07/22/99 14:55:44
6.345 07/22/99 14:59:15
7.047 07/22/99 15:01:12
7.749 07/22/99 15:03:08
8.451 07/22/99 15:05:04
9.153 07/22/99 15:07:02
9.855 07/22/99 15:08:58
10.557 07/22/99 15:10:54
Time
11.259 07/22/99 15:12:51
Time (s)
07/22/99 15:14:47
11.961
07/22/99 15:21:38
12.663
07/22/99 15:23:34
13.365
07/22/99 15:25:31
14.067 07/22/99 15:27:27
14.769 07/22/99 15:29:23
15.471 07/22/99 15:31:19
Fig. 18.26. Trip of two pumps, comparison of head variations

16.173 07/22/99 15:33:16
16.875 07/22/99 15:35:12
Brown’s Point/two pump trip/solo main/0.0008 ¥ 0.2 ¥ 0.2

17.577 07/22/99 15:37:15
18.279 07/22/99 15:39:12
18.981 07/22/99 15:41:08
Fig. 18.25. Recorded pressure variations during pump start/stop operations
19.683 07/22/99 15:43:08

20.385 07/22/99 15:45:04
bar (g)
PS head
21.087 07/22/99 15:47:01
07/22/99 15:48:57
Air and gas
flow and with compression of the gas mass, pressure rises increasingly
the gas pocket becomes smaller and the hydraulic gradient accelerating
The gas mass is still not sufficient to effectively cushion the reversed
flow persists for longer, allowing a larger reversed velocity to develop.
349
Brown’s Point/two pump trip/two mains in use/0.001 ¥ 0.2 ¥ 0.2

70
PS
60
50
Head (mAOD)
40
30
20
10
0
0.027
0.675
1.323
1.971
2.619
3.267
3.915
4.563
5.211
5.859
6.507
7.155
7.803
8.451
9.099
9.747
10.395
11.043
11.691
12.339
12.987
13.635
14.283
14.931
15.579
16.227
16.875
17.523
18.171
18.819
19.467
Time (s)
Fig. 18.27. Trip of two pumps, comparison of head variations
steeply to produce the peak pressure. With a large gas pocket there is
ample time to create the maximum reversed velocity. Eventually the
hydraulic gradient flattens and flow starts to decelerate. Gas volume
remains relatively large and rate of pressure rise is modest, producing
a gradual deceleration of flow and a lower peak. With such relatively
Brown’s Point/two pump trip/two mains in use/0.004 ¥ 0.2 ¥ 0.2

120
head – PS mid-pt
100
80
Head (mAOD)
60
40
20
0
0.027
0.675
1.323
1.971
2.619
3.267
3.915
4.563
5.211
5.859
6.507
7.155
7.803
8.451
9.099
9.747
10.395
11.043
11.691
12.339
12.987
13.635
14.283
14.931
15.579
16.227
16.875
17.523
18.171
18.819
19.467
–20
Time (s)
Fig. 18.28. Predicted head variations for increased gas release
350
Air and gas
large volumes, behaviour is similar to that which occurs in the presence

of a pressure vessel.
In these instances where the amounts of gas present are uncertain it is
prudent to consider the effects of this uncertainty by exploring sensitivity
of predictions to possible variations in gas volume. This is important as
these computations are approximate given the uncertainties surrounding
amounts of gas evolution and its subsequent behaviour.
18.8 Pump blockage

A further instance where air in a pipeline can lead to high transient
pressures is when a pump blockage occurs. Recorded pressure versus
time histories in the riser of the submersible pump installation consid-
ered earlier in this chapter revealed peak transient pressures reaching
20 bar(g). These pressures were neither associated with pump start
nor with pump trip, but occurred during pump operation. It was
surmised that temporary blockage of the operating pump could be
responsible, as considerable quantities of sediment and other material
entered the wet well after system commissioning.
Computations were carried out to determine if pump blockage could
account for the high pressures observed. The blockage was represented
as a rapid drop in pump speed with the impeller ceasing to rotate.
Pressure in the riser also decreased sharply, allowing flow into the down-
stream system to stop and the non-return valve to close. Flow passages
within the pump allowed sewage to drain from the riser, with the air
valve upstream of the check valve opening to allow air inflow. The
rate at which the riser drained back into the wet well and also the dura-
tion of the blockage could obviously vary from one occasion to another.
To illustrate events, a short duration of blockage lasting 2 s was chosen
with a reversed velocity into the wet well through the stationary pump
impeller of around 0.3 m/s.
The blockage was assumed to clear after 2 s, allowing the pump to
quickly accelerate to its design speed.
Figure 18.29 shows the predicted head variation at the pump and at
the top of the riser. After the blockage occurs, head at the pump
delivery falls towards the prevailing sewage level in the wet well. In
the riser a steep drop in head occurs and this is stabilised by operation
of the air valve. Two seconds after the blockage occurred, the pump
resumed operation and head at the pump rose to a peak and then
dropped back to some extent. Head at the top of the riser increased
and a modest oscillation developed. At around 2.9 s the air valve
351
Ayr solo pump op/blockage and restart after 2 s

180
160 Delivery Riser
140
120
100
Head (mAOD)
80
60
40
20
0
0.002
0.134
0.267
0.400
0.532
0.666
0.798
0.931
1.064
1.197
1.330
1.462
1.595
1.728
1.861
1.994
2.126
2.259
2.392
2.525
2.658
2.791
2.923
3.056
3.189
3.322
3.455
3.587
3.720
3.853
3.986
4.119
–20
–40
Time (s)
Fig. 18.29. Head variations due to pump blockage
closed and velocities upstream and downstream of the air valve connec-
tion were equalised when an abrupt upsurge developed. Peak head was
around 170 mAOD. Actual measurements indicated that maximum
pressure could reach almost 20 bar(g). The resulting elastic oscillation
decays rapidly over about 1 s. Predicted velocity variations during this
event are shown in Fig. 18.30 for the pump delivery (DN 350), at
the elevated air valve (DN 500) and at the start of the rising main
Ayr solo pump op/blockage and restart
3.0
2.5
2.0
Velocity (m/s)
1.5
1.0
0.5
0
0.002
0.139
0.277
0.415
0.553
0.691
0.828
0.966
1.104
1.242
1.379
1.517
1.655
1.793
1.931
2.068
2.206
2.344
2.482
2.620
2.757
2.895
3.033
3.171
3.308
3.446
3.584
3.722
3.860
3.997
4.135
–0.5
Time (s)
Fig. 18.30. Velocity variations during pump blockage
352
Air and gas
Ayr solo pump op/blockage and restart

40
Air volume
35
30
25
Air volume (l)
20
15
10
0
0.002
0.136
0.271
0.405
0.539
0.674
0.808
0.943
1.077
1.212
1.346
1.481
1.615
1.750
1.884
2.019
2.153
2.287
2.422
2.556
2.691
2.825
2.960
3.094
3.229
3.363
3.498
3.632
3.767
3.901
4.036
–5
Time (s)
Fig. 18.31. Air volume in riser during blockage event
(DN). Maximum velocity at the pump delivery reached 2.5 m/s.

Changing air volume in the pipeline is shown in Fig. 18.31. After the
blockage, the riser started to drain, air volume increased steadily to a
peak of around 35 litres over the 2 s interval of the blockage. After
the pump started to function, once more the air volume was vented
through the air valve over a time of around 0.2 s.
Fig. 18.32. Damaged air valve components
353
The peak pressures recorded were sufficient to damage air valves and
pressure transducers. Initially 10 bar(g) rated transducers were installed
and destroyed by the pressure transients. Replacement instruments
were rated to 20 bar(g) and survived. Air valve floats were crushed
and Fig. 18.32 shows the buckled elevator between floats of an APEX
valve typical of the damage caused.
18.9 Pumped outfall pipeline

Common features of many outfall systems are a relatively flat landward
stretch of pipeline followed by a descending seaward section leading to a
diffuser. Quite often the connection between these parts occurs at a
seawall where pipeline elevation increases locally. The relatively low
head under which these schemes operate makes them prone to the
development of sub-atmospheric pressures with the attendant riak of
gas release and formation of vapour cavities.
18.9.1 Pipeline configuration

The case considered had some additional features which made develop-
ment of cavities more likely. Figure 18.33 shows a schematic of the
75 mm recirculation
line with valve
Elevated section of 300 mm line
Tied bend
Viking Johnson coupling
Duck-foot bend Seawall with tee

M for air valve
300 ¥ 450 joint DN 450 DI

landward main
NRV
300 mm valve
Pump
Strainer
Crossings below drainage channels, DN 450 seaward
water main and main road steel main
Treated effluent lagoon
Diffuser
Fig. 18.33. Schematic of pumped outfall
354
Air and gas
Steel straps
Pipe support gantry 300∆ welded steel pipe
Viking Johnson coupling
Ground level
450∆ DI pipeline
300 ND duck-foot bend
Concrete footing
300 ¥ 450 concentric taper
Fig. 18.34. Detail of Viking—Johnson coupling
pumped outfall arrangement. Treated chemical effluent is stored within

lagoons at an elevation reaching 8.0 mAD. A strainer is provided at the
intake before flow passes through a Uniglide pump. Discharges through
the outfall are confined to a few hours around high water in the estuary,
and during the times when outflows are not permitted, flow is circulated
through the pump and back to the storage ponds through a 75 mm line
complete with a shutoff valve. Downstream of this 75 mm recirculation
branch connection is a 300 mm valve which controls discharge to the
outfall. The 300 mm welded steel pipeline is carried across the site on
gantries at a level 5.5 m above ground for most of the elevated section
but rising to 6.7 m above ground for a short final stretch. A 908 welded
joint, tied to the gantry using steel straps, is followed by a vertical
300 mm pipe containing a Viking Johnson coupling (Fig. 18.34).
Below this coupling is a 908 duck-foot bend followed by the connection
to the 450 mm landward section of pipeline. This pipeline descends
gradually from an upstream invert level of 5.89 mAD to a minimum
355
of 4.63 mAD over a length of 3450 m. At four locations along the

landward pipeline the invert falls to pass below drainage channels, a
main road and a water main. A NRV is located at chainage 3450 m.
Tide levels in the estuary can range from a minimum of 6.0 mAD
to a maximum of þ7.45 mAD. The check valve will prevent seawater
entering the landward pipeline at high tide levels. As it passes through
the seawall, pipe level rises by a modest amount to almost 5.5 mAD.
The steel seaward pipeline descends uniformly to 6.87 mAD over a
distance of 735 m and then runs more or less horizontally for a further
732 m to reach the diffuser section.
18.9.2 Viking-Johnson coupling failure

During pump operation following a period of low tide level, the Viking-
Johnson coupling in the vertical 300 mm pipeline failed. Hydraulic
transient studies were carried out in an attempt to establish the cause
of failure.
Initial interest was on conditions within the pipeline prior to pump
start. On completion of the previous pumping cycle the 300 mm
valve was closed and the pump shut down. The result is a downsurge
which produces severe vacuum pressures over most of the system.
Figure 18.35 shows head variations at the 908 tied bend and at
chainages of 1600 and 3200 m along the landward main over an 8 s
interval. Effectively the downsurge is terminated by the development
of cavities. The modest head rise at chainages 1600 m and 3200 m is
produced on flow reversal assuming the check valve at the seawall
does not close immediately on flow reversal. Figure 18.36 depicts
the envelope curves of maximum and minimum head for this event
over the 4917 m long outfall. If the NRV shuts tightly and prevents
reversed flow into the landward main and the 300 mm valve
downstream of the pump is also closed, a ‘locked-in’ sub-atmospheric
condition will persist. Only if the NRV is not drop tight or if there
are leaks along the main will any pressure relief occur. It appears
likely that at pump restart significant quantities of gas and vapour
will be present over the elevated section of pipeline and also within
the 450 mm landward main.
18.9.3 Hydrodynamic forces

Pipe joints were designed to resist a maximum pressure of 7 bar(g) and
this limit would not be exceeded were pump start to occur with a fully
356
Air and gas
70
Tied joint
Ch –1600 m
60 Ch –3200 m
50
40
Head (mAD)
30
20
10
–10
Time (s)
Fig. 18.35. Head variations after pump trip over 8 s
70
Invert level
Max. head
60 Min. head
50
40
Level (mAD)
30
20
10
–10
–20
Chainage (m)
Fig. 18.36. Envelope curves along 4.9 km long outfall
357
180
Tied joint
160 Ch –1600 m
Ch –3200 m
140
120
Head (mAD)
100
80
60
40
20
0
Time (s)
Fig. 18.37. Head variations following pump restart
primed main. In general, design of a pipe bend would be based upon the
equation:
Resultant thrust ðNÞ ¼ 2gA½h þ V 2 =ð2gÞ sinð=2Þ ð18:4Þ
where is the total angle of turn and h is the pressure head (metres). It
may be necessary to add other forces such as the weight of a water
column in some instances. The pressure head and velocity should be
the maximum occurring during the surge event.
If gas and vapour are present initially then following pump start rising
pressure will cause vapour cavities to collapse, gas to be compressed and
progressively transported and reabsorbed into the flow. Pump start
studies were carried out using a wide range of cavity volumes and
other parameters such as the rate at which gas could be reabsorbed
by the flow. It was found only too easy to create pressures substantially
greater than the 7 bar(g) design limit. Figure 18.37 shows typical time
histories of head at the tied joint and along the main. The most
likely cause of failure was considered to be movement of the 908
bend allowing the upper section of pipe at the Viking-Johnson coupling
to be pulled out. The coupling was designed to accommodate normal
thermal movements of up to possible 10 mm. It would seem likely
that the 908 had been subject to shock pressures above the 7 bar(g)
limit on various occasions when the pump was started, leading to
eventual failure of the coupling.
358
19
Relief valves
Just as underpressure can be alleviated through the supply of an additional

source of fluid to a pipeline, so an overpressure condition can be
suppressed by releasing fluid from the network either into an off-line
storage facility or from the system entirely. Compression waves generated
by pump start or by a valve closure can produce acceleration of flow in a
pipeline such that transient pressures exceed the design limit of the pipe-
line. One option for reducing maximum pressures is to have a valve which
opens automatically, discharging a sufficient amount of liquid so that the
downstream pipeline experiences a reduced rate of flow change and a
correspondingly lower surge pressure rise. A variety of pressure relief
devices are available which can be chosen to suit particular requirements.
Pressure relief valves may be attractive to the engineer faced with the
task of reducing waterhammer. Installing a device which automatically
discharges a certain flow when pipeline pressure exceeds its normal
maximum value seems an effective means of protecting the pipeline
against overpressure. If a pressure transient causing a velocity change
V travels along a pipeline of area A, it will produce a head change
a V=g. Releasing a part of the liquid media, usually to atmosphere,
at a rate Q, means that components of the pipeline system will not
be subject to the total change of flow. Instead, the velocity change
will be V—Q=A and the head change a=gðV—Q=AÞ.
To achieve this objective a number of types of relief valve are available
including specialised surge anticipation valves fitted within a pumping station.
19.1 Relief valve types

Relief valves may be installed at any location where there is a risk of
transient pressures exceeding allowable values and may be located for
359
example at a low point on a pipeline where static head is high. Alterna-

tive valve patterns available include spring-loaded types, pilot valve
operated models and instrument operated versions.
Spring-loaded versions are relatively inexpensive and less complicated
than other options. They are usually highly reliable if maintained regularly
and are available in a range of sizes, materials and pressure ratings. The
spring compression is adjustable to create the necessary closure force to
withstand internal pressure up to the set pressure at which the hydro-
dynamic force acting on the valve door element just starts to overcome
the spring force holding the valve door against its seat. A range of springs,
in either carbon or stainless steel, may be provided for each valve to give a
wide range of application. Each of the available springs would normally be
capable of adjustment to cover a range of set pressure.
There are many differences between relief valves, and Fig. 19.1
illustrates the main components of a Neyrtec relief valve produced by
Alstom. Figure 19.2 shows a typical installation for a valve of this
type. The valve will be set, usually at the factory, to open when internal
pipeline pressure exceeds an allowable value. This ‘set pressure’ may be
Deflector hood
Upper plate
Spring
Stop pieces limiting

Annular plate lateral travel
Isolating valve
From pipeline
Fig. 19.1. Elements of a relief valve
360
Relief valves
Ground level Pressure relief valve
Isolating
valve
Connection
Outflow
Pipeline
Fig. 19.2. Simple relief valve installation
5% or so higher than the maximum normal operating pressure. As

pressure rises, the valve should open fully.
Valves should operate quietly with no oscillation or hammering. It is
important that the discharge capacity of the valve is not too large for
the application. If the discharge through the valve is insufficient to
keep the valve fully open then hammering and oscillation can occur.
Noises including ‘chatter’ and ‘squeal’ have been reported when a
relief valve is oscillating. Some valves will only open completely with
a pressure 10—25% higher than the set pressure and this may be seen
as a disadvantage.
The possibility of resonant behaviour of relief valves has been the
subject of investigations over the years and manufacturers have
developed valves which seek to eliminate this risk through use of
damping mechanisms or other means such as a hydraulic shock absorber
or accumulator positioned in parallel nearby, which alters the response
of the pipeline.
Figure 19.3 shows a relief valve fitted with a spring but also having a
piston and oil-filled cylinder included. The cylinder is connected at top
and bottom to an oil reservoir which is part-filled with oil. The lower
connection includes a regulating valve which allows oil to flow from the
cylinder to the oil reservoir at a controlled rate. All other flows between
the cylinder and the reservoir occur without restriction. When pipeline
pressure exceeds set pressure, the valve opens quickly, allowing pressure
361
Oil reservoir
Spring
Cylinder
Piston
Discharge to
wet well
Valve giving control on outflow
from lower side of cylinder
From pipeline
Fig. 19.3. ‘Damped’ relief valve
relief to occur. As the valve starts to close, the restricted outflow of oil
from the lower part of the cylinder only permits a slow closure. This
damping effect also eliminates any risk of resonance developing.
Where damping is not included in the valve then it is necessary for
the period of oscillation of the valve’s spring/mass system to be much
less than that of the pipeline in order to avoid occurrence of pressure
pulsations. The moving parts should be as light as possible to reduce
inertia and ensure freedom of movement. Lightness is also favourable
for stable operation.
19.2 Initial valve sizing

Preliminary sizing of a relief valve will normally be confirmed by
computer investigation of hydraulic transient behaviour. The valve
should be sized to operate fully open.
The discharge through the valve might be based on the following
initial value for a change in flow Q:
If the time of flow change is ¼ n2L=a then valve discharge should
be set ¼ Q=n, where a is acoustic wavespeed and L is pipeline
length from source of transient to the relief valve.
362
Relief valves
The connection from the main to the relief valve should be as short as
possible to reduce any risk of resonance and to minimise head loss.
19.3 Valve positioning

As well as being found at pumping stations and on pressure vessels, relief
valves can be installed at critical locations along a pipeline and these may
be relatively remote. As with air valves, inspection, maintenance and
reliability are important. Where moving parts have to be guided there is
always the possibility that a valve will seize if elements become jammed
or encrusted with deposits. After some months of normal operation a
safety valve may not operate as intended should a waterhammer incident
occur. The factor of safety which was supposed to exist, based on analysis,
may only be an illusion under these conditions. Then what should have
been an incident may well become an accident. Valves should as far as
possible be tamperproof and provided with padlocks.
19.4 Analysis of behaviour

When a transient pressure rise arrives at the valve connection, the
valve will remain shut until pressure force from within the pipeline
Hmax = maximum set pressure head

Hmin = minimum set pressure head
Spring coils touching
Hmax Essentially straight line relationship Parabolic curve

Pressure head (m)
Hmin
Discharge (l/s)
Fig. 19.4. Head plotted against flow relationship for relief valve
363
exceeds that force exerted by the compressed spring. When the valve
opens fully (Fig. 19.4), outflow Qr is given by:
p
Qr ¼ constant 2 ðH zÞ ð19:1Þ
where Qr is discharge through the relief valve, is valve diameter, H is
head at the valve connection and z is the valve elevation. Also the
upstream and downstream characteristics (Fig. 19.5) yield the following
equations for the quasi-invariants:
Vu=s þ g=aH ¼ Jþ and Vd=s g=aH ¼ J
From conservation of volume:

Vu=s ¼ Vd=s þ Qr =A
Wavefront
Hu/s
Hset
Relief valve
Hmax
Qr
Vu/s Vd/s
J+ = Vu/s + g/aHu/s J– = Vd/s – g/aHd/s
C+ C–
Dt
Dx Dx
Fig. 19.5. Schematic for relief valve analysis
364
Relief valves
A being pipe cross-sectional area. Eliminating Vu=s and Vd=s ,

p
Jþ J 2g=aH constant 2 =A ðH zÞ ¼ 0 ð19:2Þ
Equation 19.2 can be solved for H and by substituting in the equations
for the quasi-invariants Vu=s and Vd=s are found.
19.5 Automatic surge control valve

This valve is pilot valve operated and Fig. 19.6 shows a possible
installation downstream of pumps. When pumping, the valve remains
shut. After pump trip, piezometric level downstream of the pumping
station falls and the pressure wave propagates along the main, reducing
pressure. Using a pressure sensing tapping, the pressure sensing valve
starts to open at some stage when pressure downstream of the
pump(s) has fallen below the normal minimum operating pressure.
The valve should not start to open until pressure has fallen substantially
and till velocity is small or about to reverse. If the valve is opened while
pressure and flow are still appreciable then the open valve will allow
flow to pass through the valve and allow pressure to fall more sharply
than otherwise. Once flow has reversed and pressure starts to rise
at the pumping station, the open valve allows flow to discharge to
atmosphere or possibly back into a wet well. If maximum reversed
Isolating valve Pump
Wet well
To wet well
Check valve
Pressure sensing line Isolating valve
To atmosphere
Isolating valve To rising main
Pilot operated pressure sensing valve
Fig. 19.6. Pressure sensing valve arrangement
365
flow is represented by a negative velocity Vo , then without the valve

arrangement maximum head rise will be aVo =g as flow is brought to
rest against the closed pump check valve.
With the valve open, flow will continue in the reversed direction
until the valve is closed. When the valve is set to close at a controlled
rate, deceleration of the reversed flow is gradual, thus limiting jdV=dtj
and avoiding further significant transient effects.
It will be noted that this special valve does not alleviate the initial
pressure transient or any low pressures associated with this first down-
surge and only acts to limit maximum pressures following flow reversal.
19.6 Surge anticipation valve

This is similar in operation to the automatic surge control valve but may
also open should maximum pressure exceed the normal maximum oper-
ating pressure by a preset amount. Determination of the appropriate size
of valve or valves to be used and the necessary settings is best done in
collaboration with the valve manufacturer as choice of valve size has to
be quite accurate.
19.7 Pumping station pressure relief valve

A case study involving possible use of a pressure relief valve at a pumping
station concerned a pipeline forming part of the potable water distribution
network for the State of Kuwait. A DN 800 DI rising main conveys water
from Az Zour Desalination plant to Wafra storage complex, a distance of
almost 37 km. Steady pumping pressure at Az Zour reaches 22.4 bar(g)
with the maximum allowable transient pipeline pressure being
27.5 bar(g). The pipeline is fitted with a number of Vent-o-matic air
valves which allow a relatively large air inflow while restricting the air
outflow rate. Steady pumping flow could reach 850 litres/s.
Simultaneous trip of two duty pumps at Az Zour produced the varia-
tions of head shown in Fig. 19.7, at the pumping station, at chainage
10 km and at chainage 20 km. Maximum pressure at Az Zour reached
30 bar(g). The irregular nature of head variation shown in Fig. 19.7 is
largely due to the closure of the large number of air valves which
operated during the initial downsurge which followed pumping failure.
Air valves downstream of chainage 10 km were predicted to operate.
Figure 19.8 shows the envelope curves of maximum and minimum
head along the rising main, with peak pressure remaining above
27.5 bar(g) over the first 2.3 km of main.
366
Elevation (mPWD) Head (mPWD)
0
50
100
150
200
250
300
350
0
50
100
150
200
250
300
350
0 0.112
1153.36 11.088
2367.56 22.064
3532.11 33.040
hmin
hmax
4751.58 44.016
5983.65 54.992
il (mPWD)
7155.67 65.968
8448.01 76.944
9605.88 87.920
10823.28 98.896
11979.81 109.872
13198.89 120.848
14418.39 131.823
15634.59 142.799
16858.59 153.775
18029.24 164.751
175.727
Fig. 19.8. Envelope curves following pump trip

19328.61
Time (s)
186.703
Chainage (m)
20559.99
21843.30 197.679
23005.95 208.655
219.631
Simultaneous trip of two pumps

24300.18
25532.26 230.607
26754.46 241.583
Fig. 19.7. Head variations following pump trip on a long main
252.559
28037.77
263.535
29262.87
274.511
Az Zour to Wafra rising main/trip of two pumps/Vent-o-matics
30552.27
Az Zour to Wafra rising main. Vent-o-matic air valves installed.

285.487
31841.67
296.463
33075.07
307.439
34372.03
318.415
d/s pmps
Ch. 20 km
Ch. 10 km
35607.23
329.391
Relief valves
367

Simultaneous trip to two pumps. Relief valve at PS
300
250
200
Head (mPWD)
150
100
50 d/s pmps
Ch. 10 km
Ch. 20 km
0
0.112
11.872
23.632
35.392
47.152
58.912
70.672
82.432
94.192
105.952
117.712
129.471
141.231
152.991
164.751
176.511
188.271
200.031
211.791
223.551
235.311
247.071
258.831
270.591
282.351
294.111
305.871
317.631
329.391
341.151
352.911
Time (s)
Fig. 19.9. Head variations following pump trip with pumping station relief valve

Simultaneous trip of two pumps. Pressure relief valve at the PS
300
250
200
Elevation (mPWD)
150
100
50 il (mPWD)
hmax
hmin
0
0
1153.36
2367.56
3532.11
4751.58
5983.65
7155.67
8448.01
9605.88
10823.28
11979.81
13198.89
14418.39
15634.59
16858.59
18029.24
19328.61
20559.99
21843.30
23005.95
24300.18
25532.26
26754.46
28037.77
29262.87
30552.27
31841.67
33075.07
34372.03
35607.23
Chainage (m)
Fig. 19.10. Envelope curves following pump trip with relief valve
368
Relief valves

Simultaneous trip to two pumps. Pressure relief valve at PS
250
Discharge
200
150
Discharge (l/s)
100
50
0
123.088
124.544
126.000
127.455
128.911
130.367
131.823
133.279
134.735
136.191
137.647
139.103
140.559
142.015
143.471
144.927
146.383
147.839
149.295
150.751
152.207
153.663
155.119
156.575
158.031
159.487
160.943
162.399
163.855
165.311
166.767
–50
Time (s)
Fig. 19.11. Discharge through relief valve
A pressure relief valve was included at the pumping station as a

possible option to reduce peak pressures. Simultaneous trip of two
pumps was again simulated and Fig. 19.9 shows the resulting variations
of head at the pumping station, at chainage 10 km and at chainage
20 km. The initial fall in head is the same as before but the head rise
after flow reversal only produces a peak pressure of 24.8 bar(g) at Az
Zour. The envelope curves of Fig. 19.10 show reduced maximum
pressures over about the first 15 km of main. Figure 19.11 shows the
predicted outflow through the relief valve. Two periods of operation
occurred, with the first covering an interval of 13 s and having a
maximum discharge of around 180 litres/s, while the second outflow
peaked at about 240 litres/s and had an operating time of about 4.5 s.
19.8 Grove regulator

A variation on the pressure relief valve is a regulator which employs a
rubber sleeve (Fig. 19.12). The regulator comprises a slotted core with a
central barrier. A flexible tube is fitted around the core, covering both
inlet and outlet slots. Surrounding the tube is an annular jacket space
filled with pressurised gas, usually nitrogen. When the annulus between
the outer jacket and the inner slotted cylinder is pressurised to a set
value then the rubber sleeve is pushed against the slotted cylinder,
creating an effective seal. Only when pipeline pressure exceeds the
369
Pressurised nitrogen-filled annulus

surrounding rubber tube
Slotted core Tapping
Pipeline
Inlet slots Core barrier Outlet slots

To discharge
Rubber tube
Fig. 19.12. ‘Grove’-type regulator
set pressure imposed by the pressurised annulus, is the liquid within the
system able to push the rubber sleeve aside, allowing flow to pass
through the slotted openings from the upstream to the downstream
side of the cylinder. In this way the internal pressure in the pipeline
can be limited in the same manner as achieved by the more common
pressure relief valve. Figure 19.13 depicts the sequence of operation
(a) Closed (b) Tube starting to lift
(c) Starting to throttle (d) Throttling
(e) Throttling (f) Throttle wide open
Fig. 19.13. Operational sequence of regulator
370
Relief valves
Grove Model 887 surge reliever/low and medium pressures

1600
Surge reliever flow factors (Fg)
1400 6 in. (150 mm)
8 in. (200 mm)
1200 10 in. (250 mm)
12 in. (305 mm)
Flow factor (Fg)
1000
800
600
400
200
0
0 5 10 15 20 25 30 35 40 45 50 55
% (pressure rise/set pressure)
Fig. 19.14. Flow characteristics of regulator
of the regulator. Advantages are its quiet rapid operation and positive
shutoff, even if foreign matter should become trapped between the
core and the tube. There is no risk of oscillation because of the
design of the regulator and the absence of any spring-mass system.
Discharge Qr through a regulator can be established using the
formula:
p
Qr ¼ Fg Ps ð19:3Þ
where Fg is a flow factor and Ps is the set pressure at which the regulator
is set to start opening. Valves of flow factor as a function of percentage
rise above the set pressure are shown for one model of Grove regulator
in Fig. 19.14.
19.9 High head relief valves

Some installations require a relief valve capacity much greater than
that achievable using the types of valve already described. Systems
requiring large relief valve flow rate include high dams and hydro-
power facilities.
One example of a high-performance relief valve is shown in Fig.
19.15. This valve, supplied to a number of hydroelectric installations
by English Electric Company, is of a cylindrically balanced double-
acting pattern. In the example shown, flow enters the valve through
a 710 mm diameter inlet pipe. When fully open, the valve provides a
371
372
Control shaft
Piston
(closing
Spiral member)
extension pipe Outer
cylinder
Dashpot A
C.l. of spiral casing

Cam C
Connecting pipe
Inner
body
Control valve B
Dismantling ring Bridge piece
Outer cylinder
Fig. 19.15. Cylindrically balanced relief valve

Relief valves
streamlined passage which allows flow to be deflected outwards into a

3 m long ‘hood’ of diameter 1200 mm. Water travels along the hood
in the form of a hollow jet which becomes highly aerated as it exits
from the hood. The valve and hood are angled downwards towards
the tailwater channel at an angle of 98. Energy of the flow is dissipated
by friction between the flowing water and the air and by turbulence
when the flow from the valve enters the tailwater. This downstream
channel had a length of 43 m.
In one application the cylindrically balanced valves were used to
discharge flow passing through a hydropower plant. Operating head
at the plant typically varied in the range 263—274 m depending upon
the number of turbines running. If a machine is shut down its inlet
gates are shut in around 2 s to avoid overspeed. The relief valve for
that turbine is opened in around 3 s to discharge the flow which was
previously passing through the machine. Subsequently, the valve is
closed in a time interval of 70—80 s. At full load, a turbine may consume
almost 20 t of water/s and this flow is diverted through the relief valve
when the turbine is shut down. Exit velocity was 61 m/s.
In common with many hydropower installations around the world,
the turbines were uprated with a consequent increase in the amount
of water passing through the station and during trip, through the
relief valves. Figure 19.16 shows the aerated jet exiting from the
hood and impacting with the tailwater for the uprated situation. At
low tailwater level, part of the aerated flow could be deflected upwards
Fig. 19.16. Aerated jet leaving relief valve
373
Fig. 19.17. Spillage over bridge parapet
to pass over the parapet of a road bridge (Fig. 19.17). This situation was
not considered acceptable, as one day a tourist bus stopped on the
bridge with its roof vents open. Two turbines tripped and the resulting
relief valves’ operation soaked the passengers, with water flooding out of
the bus when the door was opened. A longer tailwater channel would
have solved the problem but this was not a practical proposition. The
1200 mm diameter outlet hood was modified to include a truncated
cone containing orifices. This caused outflow to break into a large
number of jets and increased aeration and energy dissipation. While
not completely eliminating the problem of water spilling over the
parapet, some improvement has been achieved.
Other options were tried in an effort to reduce severity of the prob-
lem. One attempt adjusted the relief valve movement so that it could
only open to possibly 70%. A test using a partially opened valve setting,
and without any preliminary analysis of likely effects, produced results
described by the operators as ‘horrifying’. The entire power station
vibrated and shuddered, with penstock pressures reaching 500 psig,
which is above design maximum pipeline pressure. The relief valve
was designed to pass flow without large energy loss within the stream-
lined passages. Part-open operation creates conditions for substantial
head loss within the valve itself. Given the overall amount of energy
involved, around 40 MW/machine, any attempt to dissipate flow
energy within the valve through turbulence will inevitably result in
substantial vibration and must be avoided.
374
Relief valves
This example has illustrated that the solution does not just involve
choosing a relief valve but also involves conditions downstream of
the valve to ensure that adequate provision is made for adequate dissi-
pation of energy associated with the outflow and safe disposal of
discharging water. In the present instance, a greater length of tailwater
channel would have avoided the problems experienced.
19.10 Bursting disk

The bursting disk may also be viewed as a form of relief valve (Fig. 19.18).
Should internal pressure exceed the set pressure for operation then the
disk will ‘snap through’ and fail around its notched perimeter, allowing a
free discharge of liquid and thus alleviating rising internal pipeline
pressure. Clearly no means of shutting off the escaping flow is provided
by the failed disk and the system has to be shut down by other means.
This form of protection is relatively uncommon in water and sewage
applications.
Busting disk
Circular notch
Snap-through position
Initial position
Vu/s Vd/s
Pipeline
Dx + Dx
C+ C–
Dt
Fig. 19.18. Bursting disk
375
20
Check valve dynamics
Check valves have long been a potential source of unacceptable

pressure transients. Within pumping stations the loud bang associated
with check valve closure has been called ‘slam’. This is indicative of a
valve door or doors not closing sufficiently quickly and allowing flow
to reverse through the valve. Subsequent elimination of this reversed
flow has caused severe pressure transient effects in many installations.
Proper choice of valve can avoid development of these effects. Avoid-
ance in the first instance is much better than attempting to suppress the
effects once they have occurred.
Check valves, also called non-return valves (NRVs) or reflux valves,
are used to prevent reversal of flow. In practice with most valve types,
some flow will usually pass through the valve before it is able to close
completely. It is important to understand how a specific check valve
will response to the reflux conditions within a given pipeline system.
Two facets present themselves with regard to the task of establishing
how a check valve will respond. First there is the question of what
rates of flow change, that is accelerations or decelerations, will occur
in a given network and this aspect is discussed in the current chapter.
The second question relates to how different types and sizes of check
valve will perform within a particular system. This second aspect is
considered in Chapter 21.
20.1 Check valve response

Check valve behaviour has received a great deal of attention over the
years, not only with respect to the valves themselves but also their effect
upon pipeline systems. The relative frequency of starting and stopping of
pumps in sewage transmission systems and the resultant accumulation of
376
pressure oscillations over time has influenced recommendations regarding

allowable amplitudes of pressure transient oscillations for some types of
plastic pipes. A pumping station with a solo duty pump and with a
discharge branch size comparable to the diameter of rising main will be
more likely to produce higher amplitudes of surging along the main than
a system in which several pumps are operating and which have different
stop levels.
The ability of a check valve door to respond in a way which matches
the changes taking place in the pipeline system is of the utmost impor-
tance. The system will produce a rate of velocity change dV=dt which is
a function of the characteristics of the pipeline network and predicting
this rate of change is a primary objective of transient investigation and
the concern of this chapter.
20.2 Pumping station check valves

One of the most common applications of check valves is within
pumping stations (Fig. 20.1). A check valve is commonly sited down-
stream of each pump within its discharge branch. On occasion, a
larger check valve may also be found at the start of a rising main as
shown in the figure. This valve will provide protection for the pumping
station in the event of a pipe burst within the station. Check valves are
also found in vessel connections where they form a part of throttle
arrangements.
Within a pump discharge branch, as a check valve closes, say following
a pump failure, flow will decelerate at the valve at a rate which is depen-
dent upon a number of factors including, pumping head, flow, pump
characteristics, speed and moment of inertia. The number of pumps in
operation and the presence or otherwise of surge protection equipment
at or near to the pumping station are also important factors.
An ideal check valve would close at the instant of flow reversal but
this perfect performance is not attainable in practice. There will always
be some reverse flow in the system at the time when the valve finally
shuts. Within the same installation, alternative patterns of check
valve will respond in different ways, with the time of valve closure
being a function of the specific valve characteristics.
20.3 Consequences of an unsuitable check valve installation

When a check valve is not suitable for the installation in which it is
placed, serious consequences can result. A graphic illustration of the
377
Discharge header
Isolating valve
Isolating valve
Suction
branch
Delivery branch
Check valve
Pump
Suction header
Check valve
Isolating valve
Pressure vessel Rising main
Fig. 20.1. Check valves at a pumping station
effect resulting from inappropriate check valve selection is shown in

Fig. 20.2 which depicts the transient pressure variations measured
just downstream of the conventional swing check valve of a booster
pump serving a 30-storey apartment block (Fig. 20.3). The local
water undertaking delivered potable water to a storage facility at the
base of the building with a booster pump providing the necessary
pressure to convey water through a DN 100 wet riser to a storage
tank at the top of the structure, a height of 70 m. The pressure
378
Recorded pressure transient in

high-rise building following pump trip
350
Record from just downstream
Peak pressure of swing check valve
300 head = 350 mWG
250
Pressure head (mWG)
200
150 High-rise building
Steady running head

100 = 78.83 mWG
50
Static head Pump trip
= 75 mWG
0
0 1 2 3 4 5
Time (s)
–50
Fig. 20.2. Recorded pressure variations downstream of a check valve
Fig. 20.3. High-rise building
379
record shows steady pumping conditions before pump trip and also the
transient pressure head following pumping failure. Total period of the
record was around 5 s. Static pressure head at the check valve was
70 mWG. Steady running pressure head while pumping rose to
78.83 mWG. Subsequent to pump trip, pressure fell to a minimum of
46.61 mWG as flow reversed. Head then started to increase slightly
prior to check valve closure. After closure, transient pressure rose
abruptly to a head of 350 mWG. Elastic oscillations produced further
severe pressure peaks and after 5 s the maximum head still remained
above the height of the building. The high transient pressures caused
burst pipe connections, vibrations and noise, with the result that
occupants had to be evacuated. This is an instance of an unsuitable
check valve selection resulting in delayed valve closure and allowing
development of substantial reversed flow in the system. The remedy
was to install a valve possessing fast-closing attributes or alternatively
to use a valve with a damped closure.
20.4 Prediction of pumping station hydraulic transients

Pressure transient analysis of a complete system seeks to determine the
extreme head and flow changes which occur in the network as whole.
This will involve predicting the effects of simultaneous trip of the
maximum number of operating pumps, as well as other pump combina-
tions, in an endeavour to find maximum rates of velocity change and
peak head variations. Such an analysis would be necessary in the case
of solo pumping for the high-rise building discussed in section 20.3.
In contrast, prediction of reflux within many pumping stations requires
studying trip of one pump while the maximum number of remaining
duty pumps continues to operate. This will usually yield the maximum
rate of flow deceleration within the suction and delivery branches of the
failing pump.
Determination of the suitability of a valve for a particular application
requires consideration of two aspects.
First, prediction of the rate of deceleration and time of flow reversal
at a check valve is a matter for hydraulic transient analysis using the
techniques described in earlier chapters. This requires a model to be
set up of the pumping station and possibly part or all of the pipeline
system. It may not be necessary to model an entire network if interest
is solely in the response of the check valve. A ‘local’ model may allow
prediction of check valve closure in many instances where the closure
occurs within a relatively short time and before the entire pipeline
380
has had a chance to respond. Installations with a pressure vessel or

where several pumps are in operation are cases in which such ‘local’
models may be appropriate.
Second, the response of a particular size and pattern of check
valve, to the predicted rates of change of velocity and time of flow
reversal, requires consideration of the valve characteristics. These
characteristics can include such aspects as weight and inertia of
moving parts, travel distances for valve doors, hydrodynamic force
and/or torque acting on the valve door(s), bearing friction, and external
factors such as spring stiffness or dimensions and weights of external
levers. Some consideration will be given to specific valve types in
Chapter 21.
Some features will tend to produce more severe reflux conditions
than others and Fig. 20.1 displays a pumping station arrangement
which will tend to create a high rate of deceleration along the branch
containing a failing pump. The presence of several operating pumps
and the pressure vessel will all tend to maintain higher pressures in
the discharge manifold than otherwise. Figure 20.4 shows the hydraulic
gradient along the failing pump branch during steady pumping and also
a short time after the pump has tripped. Where there are a number of
operating pumps, the failure of one unit will cause a shift in the
operating point of the remaining pumps. The flow will not fall so
dramatically as if only one pump were in operation.
The presence of a local pressure vessel will provide a further source of
liquid in the event that a pump fails. As pressure tends to fall in the
manifold, the gas charge in the vessel expands causing outflow from
the vessel into the manifold, thus helping to sustain piezometric level
and promoting a steeper adverse hydraulic gradient along the suction
and discharge branches of the failing pump.
With piezometric level declining only slowly in the manifold, a
steep deceleration gradient is created along the branch of the failing
pump as illustrated in Fig. 20.4. As the failing pump decreases in
speed, its ability to maintain flow also decreases. Within a short
time, which can be as little as 0.2 s, the flow has reversed at the
check valve. Ideally the door of the valve should be close to its seat
at the time of flow reversal so that final closure occurs with only a
modest reversed velocity having been established during the final
stages of closure. Steepness of the deceleration gradient is also a func-
tion of the pumping head, with a higher head tending to produce a
more rapid flow deceleration for a given pumping station pipework
configuration.
381
Steady pumping head

Head loss in in manifold
discharge branch
Fall in head in vessel
due to gas expansion
Head in discharge
manifold during
pump deceleration
Head loss in vessel

connection during outflow
Adverse hydraulic gradient

producing rapid deceleration
of flow in pump branch
Pumping head during
steady operation Pressure vessel
Pumping head as M
pump decelerates
Suction well .
Discharge branch
M
Discharge manifold
NRV or header
Failing pump
Steady pumping
Suction branch hydraulic gradient
Hydraulic gradient
L during deceleration
Fig. 20.4. Hydraulic gradient in a pumping station shortly after pump trip
20.5 Reopening of a check valve door

Depending upon the characteristics of a system, such as pumping head,
static lift, etc., a check valve may reopen after initial closure. There may
be an elastic rebound of the valve door just after seating. This could
be caused by reflection of elastic waves travelling longitudinally in
the pipe wall or by reflection of pressure waves in the liquid. In addition,
as the surge progresses in the system as a whole, head downstream of the
382
check valve may fall below the upstream head in say a wet well or
suction main and forward flow may recommence with the check
valve partially reopening. A check valve installed within a bypass
around the pump(s) would also open under these circumstances. This
can occur several times before the valve finally remains shut. The
following cases illustrate both types of valve reopening.
20.5.1 Check valve reopening due to pressure wave reflections

A simple case involves a water scheme in which two pumps may operate
in parallel. Figure 20.5 shows the fairly typical layout of the pumping
Suction well +2.0 – +4.0

DN 125 suction branches
DN 150 isolating valve
DN 100 NRV
DN 150 discharge branch
Intake
WPL Duoglide
DQC 100/125
DN 125 isolating valve
DN 200 common header

8m
Surge vessel branch

Pressure vessel
at start of rising main
Fig. 20.5. Pumping station layout
383
1.2
1.0
0.8
0.6
Velocity (m/s)
0.4
0.2 Pump delivery No. 1

Pump delivery No. 2
0
0.001
0.015
0.029
0.043
0.057
0.071
0.085
0.100
0.114
0.128
0.142
0.156
0.170
0.184
0.198
0.212
0.227
0.241
0.255
0.269
0.283
0.297
0.311
0.325
0.339
0.354
0.368
0.382
0.396
0.410
–0.2
–0.4
Time (s)
Fig. 20.6. Velocities at failing/operating pump
station. Weir Pumps Limited Duoglide units type DQC 100/125 were
employed. A pressure vessel was connected 8 m downstream of the
second pump branch. Figure 20.6 shows the changing velocity just
downstream of the check valve of a failing pump and also downstream
of the valve of the remaining operating pump. After trip at 0.1 s the
velocity falls steeply and reverses at 0.285 s. A reversed velocity of
0.3 m/s is achieved at the time the check valve closes after about
0.37 s. Velocity becomes zero and remains so until pressure wave reflec-
tions cause head across the valve to allow a brief interval of reopening
before the valve finally shuts. Velocity in the operating pump branch
increases until the first check valve shuts and then falls with a
modest oscillation developing. Changing head downstream of the
check valves can be seen in Fig. 20.7, with an abrupt upsurge when
the check valve of pump No. 1 closes. Peak head is almost 120 m
above datum. Steady pumping head is around 84 mAOD. Head down-
stream of the second pump falls to a lesser extent. Wave reflection
causes head downstream of the failing pump to reach a minimum and
it is at this time that the check valve briefly reopens. A smaller head
variation occurs at the operating pump. The reason that the check
valve is able to reopen under such high head conditions is because
the pump is still rotating at around 85% of its design speed at the
time of valve closure. The reduced head developed by the pump is
still sufficient to reopen the check valve when downstream head falls.
384
140
Delivery No. 1
120 Delivery No. 2
100
Head (mAOD)
80
60
40
20
0
0.001
0.015
0.029
0.043
0.057
0.071
0.085
0.100
0.114
0.128
0.142
0.156
0.170
0.184
0.198
0.212
0.227
0.241
0.255
0.269
0.283
0.297
0.311
0.325
0.339
0.354
0.368
0.382
0.396
0.410
Time (s)
Fig. 20.7. Head at failing/operating pump delivery
As pump speed continues to decrease over time, head upstream of the

valve diminishes and the check valve remains closed.
Check valve reopening as a consequence of pressure wave reflection
is an event that will occur shortly after valve closure and can usually be
predicted by the ‘local’ type of model.
20.5.2 Valve reopening in longer term

Valve reopening in the longer term as a consequence of transient
behaviour in the system as a whole will only be predicted by a model
of the complete system, or a substantial part thereof. As an example
of this second form of behaviour, consider the case of the Abu Dhabi
sewage pumping system from pumping station No. 5 to Mafraq Waste
Water Treatment Works (WWTW), a distance of 13 343.6 m. Two
mains are used: an older pipeline and a new main. Both are of glass
reinforced plastic (GRP) and have internal diameter 1.3 m. Calcula-
tions were based upon roughness values ks ¼ 0:06 mm for the new
main and ks ¼ 0:3 mm for the existing main. A maximum of four
duty pumps may run together. These pumps are Weir Pumps Limited
SWALLOWGLIDE type SRCz 450, each with a design discharge of
607 litres/s and pumping head 67 m. Minimum transient pressure
head within the mains was set at 2.0 mWG. To protect the pumping
385
Abu Dhabi Pumping Station No. 5A

200
180
160
140
120
Head (mAD)
100
80
60
40
PS No. 5A
20 Ch. 7 km
0
0.327
11.118
21.909
32.700
43.491
54.282
65.073
75.864
86.655
97.446
108.237
119.029
129.820
140.610
151.401
162.192
172.983
183.774
194.565
205.356
216.147
226.937
237.728
248.519
259.310
270.101
280.892
291.683
302.473
313.264
324.055
Time (s)
Fig. 20.8. Head after pumping failure
mains against excessive negative pressure an air vessel installation was

provided at pumping station No. 5A. While pumping at the maximum
rate, the air volume within the pressure vessel was 15 m3 .
Following simultaneous failure of all four pumps, head at the pumping
station and 7 km along the new rising main declined as shown in Fig. 20.8.
Following the initial downsurge there is an interval of relatively small
head change at the pumping station and then a steady head rise towards
a maximum after flow reversal. Looking at the velocity downstream of the
pumps and at the start of the two mains (Fig. 20.9), it can be seen that
the flow from the pumps stops very quickly after pump trip and the check
valves close. This rapid response is primarily due to the presence of the
pressure vessel. After about 24 s, check valves reopen and velocity
upstream of the pressure vessel reaches almost 1.3 m/s before gradually
declining to zero at around 65 s. A bypass fitted with a check valve
would operate in a similar manner.
The second check valve closure is much more gradual than the first.
Thereafter the check valves were predicted to remain shut. The reason
for reopening of the check valves is because of the low head which has
developed by 23 s after trip (Fig. 20.8). This low head is less than that in
the suction well and this has allowed flow to become re-established
through the pumps. As flow reverses in the new main (Fig. 20.9), this
386

3.5
New main
3
Old main
PS No. 5A
2.5
1.5
Velocity (m/s)
0.5
0
0.327
5.232
10.137
15.042
19.947
24.852
29.757
34.662
39.567
44.472
49.377
54.282
59.187
64.092
68.997
73.902
78.807
83.712
88.617
93.522
98.427
103.332
108.237
113.142
118.048
122.953
127.858
132.763
137.668
142.572
–0.5
–1
–1.5
Time (s)
Fig. 20.9. Velocity after pumping failure
causes head at the pumping station to rise, with the result that the flow
through the pumps decelerates and the check valves reclose quietly.
The necessary pressure vessel capacity is reduced by reopening of the
check valves. Figure 20.10 shows air volume expanding quite rapidly

60
Vessel
50
40
Air volume (m3)
30
20
10
0
0.327
10.791
21.255
31.719
42.183
52.647
63.111
73.575
84.039
94.503
104.967
115.431
125.896
136.360
146.823
157.287
167.751
178.215
188.679
199.143
209.607
220.070
230.534
240.998
251.462
261.926
272.390
282.854
293.318
303.781
314.245
324.709
Time (s)
Fig. 20.10. Air volume after pumping failure
387
after pumps were tripped. When the check valves reopen, only a modest
additional expansion occurs, with air volume reaching a maximum of
60 m3 . Effectively, the recommencement of flow through the check
valves has replaced the need for further outflow from the pressure
vessel.
20.6 Check valve response in a multi-pump installation

In a multi-pump installation, maximum upsurge at the check valve is a
function of the number of operating pumps. Figure 20.11 shows one
example of how the maximum transient pressure varies with number
of operating pumps when one of the pumps is tripped.
1.8 1 pump tripped and remaining

pumps continue running
1.6
1.4
Reversed velocity (m/s)
1.2
1.0
0.8
0.6
0.4
0.2
1 2 3 4
Number of operating pumps
Fig. 20.11. Reversed velocity plotted against number of operating pumps
20.7 Surge behaviour as a check valve shuts

Figure 20.12 shows a sketch of transient piezometric level around a
check valve before and just after closure. The sudden head rise down-
stream of the valve following closure is accompanied by a corresponding
head drop upstream of the valve. If head upstream falls below vapour
pressure or gas release head then the minimum head will be determined
by these effects. Notwithstanding this complication, the differential
head across the closed check valve door will impose a longitudinal
388
Compression wave travelling

downstream after valve closure
dx/dt = +a
Differential head across
valve door after closure a/gVo
Rarefaction wave travelling

upstream after valve closure Transient hydraulic gradient
along failing pump branch
before NRV closure
a/gVo
dx/dt = –a Door open
Vo V=0 V=0 Vo
Door closed
Reversed velocity before closure
Fig. 20.12. Head conditions when check valve shuts
thrust on pipework. Deceleration of the door itself as it meets its seat

will add to the thrust. The pipe, pump and valve supports should
together be able to transfer this thrust to the foundations of the station
without incurring unacceptable movement. Failure to absorb this force
may result in the pump and valve being pushed out of position and
cracking where the suction branch passes through the sump wall.
Noise and vibration often accompanies check valve closure.
For the situation shown in Fig. 20.12, the differential head across the
check valve produces a hydrodynamic thrust in the upstream direction:
Thrust ¼ 2aAVo ð20:1Þ
The time of closure of a free-acting check valve is strongly dependent
upon the response of the system within which it is installed. Not all
computational models will be able to integrate movement of the
check valve with the hydraulic transient analysis in order to establish
times of check valve closure. In carrying out calculations using a
model of this type where a check valve is involved, a time of valve
closure should not be imposed as with an actuated valve. To pre-
select a time of closure without reference to the system response on
389
which check valve closure depends, risks numerical prediction of surge

effects which will not actually occur.
A two-stage process can be used to determine an appropriate valve
type. First the model can be used to simulate pump failure ignoring
the influence of the check valve. This will yield the variation of
dV=dt at the point where the valve is to be placed. Using charts for
individual valve types (see Chapter 21), relating dV=dt to reversed
velocity at the moment of closure, then an appropriate reversed velocity
and closure time can be determined for a candidate valve. The values of
closure time or reversed velocity can now be used in a rerun of the
model to predict transient effects within the system after closure.
Some charts showing valve closure performance as a function of
dV=dt have been included in Chapter 21.
20.8 Modelling a pumping station

The approach to modelling may be dependent on the capabilities of the
computer model available. If the modelling exercise is only one part of an
overall study of hydraulic transients in a system then the computer model
may reasonably be expected to encompass the entire network. As pipe
lengths within a pumping station may be quite short, this implies a
small time increment which may not be particularly suitable for the
much larger pipe lengths appropriate to the majority of the network. A
modified computational scheme using two time step sizes and which
was described in Chapter 6 may be useful in this context. The author
has used this technique on a number of occasions in these circumstances.
If interest is centred on behaviour of the check valve alone and
transient behaviour in the system as a whole does not require to be
modelled then it may be possible to terminate the computer model
local to the pumping station using a non-reflecting boundary, also as
described in Chapter 6. This can be done when it is considered that
the check valve will close in a short time relative to the wave reflection
time from the nearest feature along the pipeline, which will produce a
significant response. In this instance a non-reflecting boundary can be
introduced close to the pumping station.
20.8.1 Non-reflecting boundary with allowance for external pipeline

resistance
The technique as described in Chapter 6 is clearly approximate since
although no effects will return to the boundary from discrete features
390
+t
V, H
C Co
Dt
T/2
Dt Vs, HT/2
Dt
Steady-state region
V = Vs (constant)
T/2
Dt
Cs
t = 0.0 (start of computation)

Dx Dx
Vs, Hs
Steady flow values
+s
Part of system External network

being modelled
Fig. 20.13. Non-reflecting boundary with resistance allowance
along the pipeline such as changes of cross-section, bifurcations, valves,

etc. there will be a more gradual and continuous change due to the
action of pipeline resistance. If necessary, some allowance can be
made for this effect by slowly modifying the invariant value from the
external system over time. Allowance for resistance within this external
system can be accommodated in a number of ways.
One approach is illustrated in Fig. 20.13. The sign convention for
þve flow direction uses a code number s ¼ 1. If þve flow is from
inside the modelled area towards the boundary then s ¼ þ1 otherwise
s ¼ 1. Starting from steady flow conditions at time t ¼ 0, a character-
istic Cs can be pictured leaving the boundary with initial steady flow
values Vs and Hs . As this path propagates into the external system it
defines a boundary between a region of known steady-state conditions
and an area in which potentially conditions of motion are changing with
391
time (although not necessarily at once). Along this characteristic Cs ,

while the velocity remains essentially constant, ignoring the effects of
compressibility and pipe wall deformation, head will gradually change
under the action of pipeline friction. Assuming a constant wavespeed
a, the quasi-invariant along Cs may be written as:
ð
Js ¼ Vs þ sg=aHs g S dt ð20:2aÞ
Representing friction gradient S using the D’arcy equation and

assuming a constant f value then:
S ¼ f Vs2 =ð2gDÞ
or
ð
Js ¼ Vs þ sg=aHs g f Vs2 =ð2gDÞ dt ð20:2bÞ
With reference to Fig. 20.13, over a time interval 0 to T=2, the quasi-
invariant becomes:
Js ¼ Vs þ sg=aHs f T=ð4DÞVs2
and
HT=2 ¼ Hs f T=ð4DÞVs2 ð20:3Þ
Proceeding along the characteristic Co from time T=2 to a solution
point on the boundary at time T, the quasi-invariant value is:
ð
Jo ¼ Vs sg=aHT=2 g S dt ð20:4Þ
with the integral evaluated between the limits T=2 and T. Applying the
D’arcy equation once more p and approximating average velocity, over
the interval T=2 T, by ðVVs Þ then:
S ¼ f VVs =ð2gDÞ
and at the solution point on the model boundary:
V sg=aH ¼ Jo ¼ Vs sg=a½Hs f T=ð4DÞVs2 f T=ð4DÞVVs
or rearranging,
V½1 þ f T=ð4DÞVs sg=aH ¼ Vs sg=a½Hs f T=ð4DÞVs2 ð20:5Þ
From within the region being modelled the characteristic C arrives at
the solution point with a quasi-invariant value:
V þ sg=aH ¼ J
392
Adding these simultaneous equations then,

V ¼ fVs sg=a½Hs f T=ð4DÞVs2 g=½2 þ f T=ð4DÞVs ð20:6Þ
Substitution of this velocity value into the equation for J yields the
corresponding value of head H.
In the case of a booster pump installation it would be necessary to
incorporate two non-reflecting boundaries as illustrated in Fig. 6.13
or indeed as many such boundaries as the configuration required.
Obviously use of this technique is not restricted to pumping installa-
tions but can be applied to other circumstances. Within the pumping
station model, all of the features found in a pipeline network, such as
valves, branches, pressure vessels, air valves, may be represented.
20.8.2 System curve boundary

Where a multi-pump installation is being examined it will usually be
necessary to simulate the effects of different numbers of operating
pumps. The initial conditions require consideration of the system
curve of the downstream system and in the case of a booster pumping
station the head—flow relationship of the suction main system is also
necessary. Figure 20.14 shows the boundary of that part of the system
being modelled and the system curves which apply to a rising main
and suction main. Digitising the system curve provides the necessary
auxiliary equation to permit solution on the boundary after each incre-
ment of time. Assuming a constant wave speed a, then if conditions lie
Limit of modelling
Part of system being modelled External system

+s
H System curve
d/s system
H(i+1)
i
H(i)
Dt
System curve
u/s system
Q(i) Q(i+1) Q
Dx
Fig. 20.14. Use of ‘system-curve’ as a boundary condition
393
in the segment of curve between points ‘i’ and ‘i þ 1’:

dH=dQ ðHði þ 1Þ HðiÞ Þ=ðQði þ 1Þ QðiÞ Þ ¼ H0
or
H ¼ HðiÞ þ H0 ðAV QðiÞ Þ
and from the characteristic arriving at the boundary, the quasi-invariant J
can be written:
J ¼ V þ sg=aH
Solving for velocity then:
V ¼ ðH0 QðiÞ HðiÞ þ sa=g JÞ=ðH0 A þ sa=gÞ ð20:7Þ
Head H can be found by substituting the above velocity value into the
quasi-invariant relationship at the boundary.
The non-reflecting boundary may be used in tandem with the system
curve boundary to examine the effect of trip of different numbers of
operating pumps. The system curve boundary can be employed to set
up steady flow conditions for any given number of running pumps. A
subsequent computation using the non-reflecting boundary would trip
one or more of these pumps to establish the deceleration gradient
dV=dt at a failing pump. The system curve boundary can then be
used to set up a new steady flow regime for a different pumping config-
uration before using the non-reflecting boundary once more to simulate
further pump trip events.
20.9 Reduction of transient pressures following valve

closure
Pressure waves are reflected as normal from system components such as
bifurcations, vessels, etc. Since pipe lengths in the pumping station tend
to be relatively small, the period of elastic oscillations will also be short.
These pressure waves generated at the check valve will travel to the
discharge manifold or the start of the rising main. Depending upon
the relative diameters of pump branch pipes, manifold and rising
main, appreciable reduction in amplitude of oscillations can occur as
the wave travels into these components.
Two examples will be considered: a manifold or header and an
increase in pipe cross-section.
Where a manifold or header is present having cross-sectional area
Am and the pump branch has area Ab then the reduction in pressure
394
head rise Hm at the manifold can be expressed as a proportion of the

corresponding head rise Hv at the NRV. At the instant the NRV
shuts, let the velocity in the branch be Vo . For continuity, the flows
in the manifold must balance. Initial flow can be distributed in the
manifold in an infinite number of ways. Assume all flow comes from
the upstream side with velocity Vo Ab =Am . Flow in the other part of
the manifold is then zero. A different distribution of manifold flows
or assumption about þve flow directions will produce the same result
as the following analysis.
At a closed check valve, head rise Hv when reversed velocity Vo is
eliminated has often been quoted as having a maximum value:
Hv ¼ a=gVo
This applies at the downstream valve face at initial closure and assumes
no movement of the valve longitudinally in response to thrust directed
along the axis of the pipeline. Observations have shown that it is quite
possible for maximum pressure rise in the pump branch upstream of a
check valve to exceed the value aVo =g. Discussion of circumstances
when this can occur is in section 20.10.
When the compression wave after check valve closure reaches the
manifold connection, head at the junction is given by:
X X
Hm ¼ 1=g ðA JÞ= ðA=aÞ ð6:4Þ
or assuming constant wavespeed a:
X X
Hm ¼ a=g ðA JÞ= A ð20:8Þ
Neglecting pipeline losses and substituting the appropriate invariant
values:
Hm ¼ a=gfAb ðVo Þ þ Am ðVo Ab =Am Þ 0g=fAb þ Am þ Am g
or
Hm ¼ a=gVo 2=ð1 þ 2Am =Ab Þ
Ratio of head rise at the junction to head rise at the valve face Hm =Hv
is:
Hm =Hv ¼ 2=ð1 þ 2Am =Ab Þ ð20:9Þ
In the second example, when the pump branch meets a rising main
with larger diameter, then assuming a constant wavespeed a and
neglecting pipeline losses, head rise Hv at the closed NRV is again:
Hv ¼ a=gVo
395
Head rise Hm at the start of the rising main is given by:

X X
Hm ¼ a=g ðA JÞ= A ð20:8Þ
Substituting the appropriate invariant values the head rise in the rising
main is:
Hm ¼ a=gVo fAb ðVo Þ þ Am ðVo Ab =Am Þg=fAb þ Am g
The ratio of head rise in the main to head rise at the NRV is then:
Hm =Hv ¼ 2=ð1 þ Am =Ab Þ
or in terms of relative diameters Db and Dm , the ratio is:
Hm =Hv ¼ 2=½1 þ ðDm =Db Þ2 ð20:10Þ
For example, if branch diameter is 12 rising main diameter then:
Hm =Hv ¼ 0:4
Equations (20.9) and (20.10) indicate that pressure transients
produced by delayed check valve closure can be of high amplitude
but with severity decreasing quickly as the pressure wave travels into
the network. Due to the short stretches of pipework usually associated
with pumping stations, wave reflections and relief of pressure usually
occur relatively quickly so that duration of such excessive pressures
can likewise be short.
20.10 Maximum pressures at a check valve
20.10.1 Initial valve closure

When a check valve closes, the reversed velocity which has been estab-
lished is abruptly eliminated, with a corresponding head rise at the
downstream valve door face and head drop at the upstream face of
the valve. Subsequently, pressure waves propagate in both directions
from the valve. A compression wave will travel into the pipeline
system downstream of the valve and a rarefaction wave will propagate
upstream of the valve face. Local cavitation may occur at the upstream
face of the valve.
Initial closure of a valve such as a swing check is accompanied by
abrupt deceleration of the reversed velocity Vo passing through the
valve. At the downstream face, the resulting inertial head rise Hv is
commonly taken as:
Hv ¼ aVo =g
396
The value quoted above is sometimes stated as the maximum head rise
possible at a check valve on closure. This is only the case as far as the
initial head rise is concerned, as observations have revealed consider-
ably greater head rise after initial closure as described in section
20.10.2. At the upstream valve face, the corresponding head drop
will be:
Hv ¼ aVo =g
20.10.2 Cavitation upstream of the valve and resulting peak

pressures
At the upstream valve face there is a fall in head which in the absence of
other influences will be aVo =g. Since pressure head Ho at the valve
just prior to closure may be relatively low, it is quite likely that minimum
pressure at the upstream face of the valve calculated as Ho aVo =g will
be less than i:l: hvap , where i:l: is invert level and hvap is the vapour
pressure head. Should minimum head reach the vapour pressure
head, a vapour-filled cavity will start to develop so that:
Vo þ g=aHo ¼ Jþ ¼ V þ g=aði:l: hvap Þ
Taking the cavity to be distributed across the entire pipe cross-section,
then velocity at the front of the cavity will be:
V ¼ Vo þ g=aHo g=aði:l: hvap Þ ð20:11Þ
Flow in the pipe branch upstream of the valve will decelerate and
reverse becoming positive.
The cavity will then collapse with an attendant shock pressure rise
against the upstream valve face. By this time, pressure wave effects in
the pipe branch downstream of the check valve may well have dissipated.
In his investigations of 50 mm swing check valve behaviour, Worster
(1959) observed that in almost all cases studied, after initial seating of
the valve, the door repeatedly opened by a very small amount and then
shut due to the action of pressure waves in the pipe branch upstream of
the valve. The differential transient pressure across the valve door was
sufficient to reopen the valve. In more severe tests when appreciable
noise was produced by initial closure, the valve remained shut for a
longer period and then reopened sometimes by as much as 128 from
its seat. Following closure, further periods of reopening and closure
occurred. Worster attributed this behaviour to high surge pressures
after cavity closure, producing a thrust against the valve door causing
it to reopen. Forward velocity in the entire upstream water column as
397
Velocity at pump discharge following pump trip at time = 0.1 s

3
2.5
1.5
Velocity (m/s)
0.5
0
0.001
0.109
0.216
0.323
0.431
0.538
0.645
0.753
0.860
0.967
1.075
1.182
1.289
1.397
1.504
1.611
1.718
1.826
1.933
2.040
2.148
2.255
2.362
2.470
2.577
2.684
2.792
2.899
3.006
3.114
3.221
–0.5
Time (s)
–1
–1.5
Fig. 20.15. Velocity coast-down after pump trip
the cavity collapsed was of a similar order as the initial reversed velocity
at the time of initial closure.
Miller (c. 1970) observed behaviour of a Glenfield and Kennedy
Swing Check Valve subsequent to pump trip from an initial steady
pumping head of 14.33 m. Following check valve closure, a peak pres-
sure head of 405 ft or 123.44 mWG was measured at the inlet to the
check valve while the corresponding peak pressure at the valve outlet
was 270 ft or 82.3 mWG. Figure 20.15 shows predicted coastdown of
velocity at the pump discharge. After closure, the velocity oscillates
as a vapour cavity upstream of the check valve repeatedly opens and
shuts. The corresponding pressure head is shown in Fig. 20.16 together
with the observed values. A more detailed picture of head changes can
be seen in Fig. 20.17. It will be noted that head variations downstream
of the valve occur with higher frequency and dissipate more rapidly.
Modest vapour cavities were predicted to occur on the downstream
side of the valve but these were smaller and relatively short-lived
under the generally higher pressure prevailing in the downstream pipe-
line. The head changes upstream of the valve take place over a longer
time span with larger cavities forming. From this and other studies it has
been noted that transient pressures upstream of a check valve can
exceed the corresponding downstream pressures often by a considerable
margin. Changes in diameter along the pump suction and discharge
branches can aggravate conditions.
398
Pressure head at inlet and outlet of Glenfield and Kennedy Swing Check Valve
following pump trip at time = 0.1 s
160
140
Valve inlet
123.44 metres – peak head
120 measured at valve inlet
Pressure head (mWG)
Valve outlet
100
82.3 metres – peak head
80 measured at valve outlet
60
40
20
0
0.001
0.121
0.240
0.359
0.478
0.597
0.716
0.835
0.954
1.073
1.192
1.311
1.430
1.549
1.668
1.788
1.907
2.026
2.145
2.264
2.383
2.502
2.620
2.700
2.858
2.977
3.096
3.216
3.394
3.4
3.5
–20
Time (s)
Fig. 20.16. Head plotted against time as check valve shuts
The behaviour described is another instance of a cavity allowing

time for an appreciable velocity to develop whose subsequent rapid
deceleration leads to high transient pressure. Any type of cavity,
whether air-filled, gas-filled or vapour-filled, may provide time for an
Detail of pressure head at inlet and outlet of Glenfield and Kennedy

Swing Check Valve following pump trip at time = 0.1 s
160
Valve inlet
140
120
Valve outlet
Pressure head (mWG)
100
80
60
40
20
0
2.570
2.596
2.623
2.649
2.675
2.792
2.728
2.755
2.781
2.808
2.834
2.861
2.887
2.914
2.940
2.967
2.993
3.019
3.046
3.072
3.099
3.125
3.152
3.178
3.205
3.231
3.258
3.264
3.311
3.367
3.383
–20
Time (s)
Fig. 20.17. Head plotted against time as check valve shuts (detail)
399
excessive velocity to develop with consequent pressure rise. Physical

characteristics of cavities may differ depending upon their content.
Water vapour will recondense as pressure rises above vapour pressure
with virtually no back-pressure to decelerate the developing velocity
whereas an air-filled cavity will tend to impose some deceleration on
flow as the air becomes compressed.
20.11 Other applications of check valves
20.11.1 Check valve at the start of a rising main

Check valves may be located at the start of a rising main to prevent
flooding of a pumping station in the event of a pipe burst either
within or close to the station. If a pressure vessel is present, this
check valve should be sited upstream of the vessel connection to
avoid the vessel becoming isolated from the main following valve
closure. This type of installation is more usual in higher head systems
where the effects of a discharge into the pumping station could be
more severe. A rising main check valve will usually be at least the
same diameter, and often larger, than the pump discharge check
valve. A larger valve of the same pattern in the rising main will close
more slowly than the valve at the pump discharge in the event of
pump trip but should a pipe burst occur just upstream of the large
check valve a more rapid closure will occur. Hydraulic transient
investigation of this event should be carried out to ensure that safe
closure can be achieved without incurring further severe surge effects
after the valve shuts. Consideration should also be given to the possi-
bility that a larger check valve of a light-door pattern, at the start of
a rising main, may oscillate if not held fully open by the flow.
20.11.2 Check valve on vessel connection

This valve will usually be required to open promptly in the event of
outflow from the vessel, tending to indicate a light-door valve. There
is evidence of valve stops being damaged on commencement of outflow
from the vessel, where a heavier door is thrown open against its stop.
20.11.3 Bypass check valve

A check valve may be required on a bypass line in a pumping station
(Fig. 20.18). The bypass arrangement only comes into play when the
400
Bypass line NRV
Pump
Suction header Discharge header
NRV
Fig. 20.18. Check valve on bypass line
head, or piezometric level, downstream of the pump(s) falls below the

suction head. The bypass check valve then opens, allowing flow to
pass through the bypass around the pumping station without having
to pass through the pump. Without this arrangement the pump will
continue to rotate, absorbing energy. This type of arrangement is
particularly useful if for example the pumps are acting to augment
gravity flow.
When pumps are tripped, piezometric level downstream of the pumps
falls below suction level and flow commences in the bypass. The check
valve should be able to open without undue delay but more importantly
should be able to close promptly in the event of pumps being restarted,
otherwise slam could occur when the check valve door finally closes and
reversed flow in the bypass is eliminated.
20.11.4 Check valves along a rising main

A further interesting application of the check valve is as a pressure
transient control device placed along a rising main. Consider a pumping
main operating at intermediate to high head, such as may produce
401
relatively high adverse gradients along the pipeline after a pumping

failure. Such gradients may lead to high upsurge pressures as the
reversed flow is developed.
When a pressure vessel is to be refilled by the reversed flow, then
introducing a check valve or valves along the main prevents refilling
of the vessel to the extent which would be imposed by the full static
head. This may mean that the need for throttling in the vessel con-
nection is reduced or removed entirely. If no bypass is included at the
in-line check valve, then the vessel gas volume will remain close to
the maximum expanded volume and will not refill.
Choosing a check valve for this application may not necessarily
require a fast-closing performance. Depending upon the size of main,
a swing-check valve or a split-disk pattern with light springs may be
used. A valve pattern involving a sliding movement of the door
under the action of an axial spring has also been applied. The use of
check valves in this type of application should arrange for the valve
to open fully to avoid risk of ‘chattering’. The extra head loss associated
with the in-line application of the valve(s) should also be considered
when compiling system curves and selecting pumps. It may be necessary
for an in-line check valve to be in parallel with an isolating valve of
similar diameter to allow the check valve to be removed for periodic
maintenance.
Consider the case of the raw water transfer system from the River
Nile at Gizara to 6th October City. The pumping system is in two
parts. The first stage involves an intake and pumping station on the
River Nile at El Dahab. Discharge is into a DN 1500 DI main of
length 16.5 km. Flow from the main enters storage tanks before being
pumped onwards by the booster pumping station which is the start of
the second stage of the project. Flow then travels a further distance
of 10.25 km through twin DN 1000 DI mains to the treatment works
serving 6th October City. Design flow rate is 2400 litres/s with three
duty pumps in service at El Dahab.
The first-stage pipeline commences at an elevation of þ15 mASL
and runs at relatively low level for 9 km before climbing steeply for
around 3 km to reach a general elevation of þ100 mASL. Thereafter
the line crosses a plateau at a level of þ90 mASL to þ100 mASL
before reaching storage facilities at the booster station. The line is
fitted with three NRVs at chainages 8850 m, 9750 m and 10700 m.
These NRVs are of the torsional spring-assisted closure, split-disk type.
A pressure vessel was included at the pumping station with a steady
pumping air volume of around 20 m3 . In predicting pressure transient
402
Gizera/El Dahab pumping station and DN 1500 rising main

120
100
80
Elevation (mASL)
60
40
i.l. mASL
20 hmax
hmin
0
0
575.6
1247.1
1726.7
2398.3
3069.8
3645.3
4316.9
4892.4
5468.0
6043.6
6619.2
7098.8
7578.5
8154.1
8633.7
9209.3
9784.9
10 360.5
10 840.1
11 415.7
12 087.2
12 662.8
13 334.3
13 909.9
14 389.5
14 965.1
15 540.7
16 212.2
Chainage (m)
Fig. 20.19. Envelope curves after pumping failure
behaviour in the entire system, simultaneous failure of three duty

pumps was examined with pipeline resistance at its minimum assessed
value. Figure 20.19 depicts envelope curves for this event. Minimum
hydraulic level falls relatively smoothly until the first NRV where
there is an abrupt head rise of around 50 m. Thereafter more modest
upward steps in minimum head occur at the second and third NRVs.
Maximum head falls slowly over the upstream 10.7 km and then
there is a modest step rise across the third NRV.
Looking at the predicted velocity variations at the pumping station
downstream of the vessel and at the three NRVs (Fig. 20.20), it can
be seen that flow reversal and check valve closure at the pumping
station occurs very shortly after trip at 1 s. Downstream of the pressure
vessel a much more modest deceleration takes place with flow reversal
after 34 s. Deceleration at the check valves has the same form as
downstream of the vessel and the valves shut after about 37 s with
only a small time difference between valve closure times. The curves
of Fig. 20.20 are important in providing an accurate assessment of
deceleration rates at these valves, thus allowing an appropriate type
of valve to be chosen. In this case the rates of flow change are such
that a high closure performance is not required.
403

1.4
PS
1.2 d/s vessel
u/s 1 NRV
1 u/s 2 NRV
u/s 3 NRV
0.8
Velocity (m/s)
0.6
0.4
0.2
0
0.093
2.139
4.185
6.231
8.277
10.323
12.369
14.415
16.461
18.507
20.553
22.599
24.645
26.691
28.737
30.783
32.829
34.875
36.921
38.967
41.013
43.059
45.105
47.151
49.197
51.243
53.289
55.335
57.381
59.427
–0.2
–0.4
Time (s)
Fig. 20.20. Velocity changes after pumping failure
Closure of the first NRV largely deprives the pressure vessel of

refilling water and Fig. 20.21 shows that after the air volume expands
following pumping failure there is only a small reversed flow and the
volume oscillates around an eventual steady volume of almost 60 m3 .

70
60
50
Air volume (m3)
40
30
20
10
Vessel
0
0.093
3.906
7.719
11.532
15.345
19.158
22.971
26.784
30.597
34.410
38.223
42.036
45.849
49.662
53.475
57.288
61.101
64.914
68.727
72.540
76.353
80.166
83.979
87.792
91.605
95.418
99.231
103.045
106.858
110.671
114.484
118.297
Time (s)
Fig. 20.21. Vessel air charge after pumping failure
404

120
Gizera
u/s 1 NRV
100 d/s 1 NRV
80
Head (mASL)
60
40
20
0
0.093
4.092
8.091
12.090
16.089
20.088
24.087
28.086
32.085
36.084
40.083
44.082
48.081
52.080
56.079
60.078
64.077
68.076
72.075
76.074
80.073
84.072
88.071
92.070
96.069
100.068
104.068
108.067
112.066
116.065
120.064
Time (s)
Fig. 20.22. Head variations after pumping failure, upstream part of system
When pumps are restarted they will commence operation against a low
static head and with the start-up compression wave being reflected from
the first closed check valve.
Figure 20.22 shows the predicted head variations at the pumping
station also upstream and downstream of the first NRV, after pumps
are tripped. A smooth fall in head at the pumping station is followed
by a very modest oscillation around an eventual static head of about
45 mASL. Upstream of the first NRV the head drop initially follows
the same form as at the pumping station until a wave reflection from
downstream produces a head rise beginning after about 14 s. A
second reflection causes head to fall once more and at around 37 s
the NRV shuts. The rate of head drop increases upstream of the
valve after closure and head falls to a minimum at about 58 s.
Thereafter an oscillation occurs between the pumping station and
the NRV, with maximum amplitude at the closed valve. Downstream
of the NRV, head conditions are essentially the same as upstream.
After closure a modest high-frequency oscillation occurs in that section
of main between the now closed first and second NRVs.
Figure 20.23 depicts changing head downstream of the second and
third NRVs. After closure of the check valves a modest high-frequency
oscillation is developed between the shut second and third valves.
405

120
100
80
Head (mASL)
60
40
20 d/s 2 NRV
d/s 3 NRV
0
0.093
4.092
8.091
12.090
16.089
20.088
24.087
28.086
32.085
36.084
40.083
44.082
48.081
52.080
56.079
60.078
64.077
68.076
72.075
76.074
80.073
84.072
88.071
92.070
96.069
100.068
104.068
108.067
112.066
116.065
120.064
Time (s)
Fig. 20.23. Head variations after pumping failure, downstream part of system
Downstream of the final NRV the oscillation is complicated by fluc-

tuations resulting from action of air valves along the final 4.5 km of
main. Positioning of the in-line check valves should seek to avoid the
possibility of vacuum pressures developing at the upstream side of the
valve following closure against a reversing flow. Valves should be
located at points on the pipeline of sufficiently low elevation to
prevent this from occurring.
20.11.5 Inclusion of air valves with in-line check valve

If the pipeline is fitted with air valves which function during the
downsurge allowing air inflow, or if a pressure vessel is installed at the
pumping station, then as flow comes to rest and starts to reverse, air
volumes will exist in the pipeline together with appreciable adverse
hydraulic gradient.
Often air valves are purchased before pressure transient analysis has
been undertaken. Where these valves are of a typical large-orifice type,
giving large inflow and outflow capacity, then a steep adverse hydraulic
gradient will potentially produce high reversed velocity and substantial
abrupt head rise at the valve when it shuts, as described in Chapter 17.
Where the opportunity exists to change the valves then restricted
406
Downstream
reservoir
Piezometric gradient
without in-line NRV
Reversed flow
Double-orifice air valve
with in-line NRV
In-line NRV with bypass

Rising main
Closed NRV
PS
Fig. 20.24. Hydraulic gradients with in-line check valve and bypass
outflow valves may be used to prevent unacceptable secondary tran-

sient effects developing when the valves close. The general principle
is to flatten the adverse hydraulic gradient to remove at least part of
the driving force causing reversed flow.
Where standard large-orifice valves, without outflow throttling, are
used then a check valve or valves can be installed at appropriate loca-
tions along the main. Then reversed flow can be avoided entirely or if a
throttled bypass is provided around the check valve, additional head
loss can be introduced (Fig. 20.24). Allowing some flow through the
bypass dissipates energy and provides an alternative means of flattening
the adverse hydraulic gradient. The extent of throttling controls the
magnitude of reversed velocity so that air valves can close without
severe secondary transient effects developing. Siting of the check
valve has to be at an elevation such that sub-atmospheric pressures
are avoided upstream of the valve.
20.11.6 Backflow check valve

Figure 16.3 shows the pipeline arrangement at a downstream receiving
reservoir which allows reversed flow out of the reservoir. The outflow
pipeline contains a check valve to ensure that flow entering the reser-
voir does so through the normal filling connection. This backflow
arrangement allows water to enter the upstream pipeline in the event
of flow reversal such as might be required to refill a pressure vessel or
407
to permit purging of air through air valves. Without this arrangement

the filling connection would empty and possibly a section of pipeline.
This would reduce piezometric level possibly below prevailing ground-
water levels. It will be noted that this back flow arrangement will
have no influence upon transient events until reversed flow has
occurred. There is generally no particular requirement for this check
valve to open rapidly or to close quickly.
408
21
Check valve characteristics
Control of pressure transients is essentially a matter of limiting rates of

velocity change. In the context of check valves, the velocity in question
is the reversed velocity in the pipeline as the valve closure element
approaches its seat. To limit severity of pressures on valve closure,
two principal alternatives are available. First, measures can be taken
to minimise the velocity at the time when the valve closes. This is
achieved by selecting a check valve which has the ability to shut quickly
so that a strong reversed velocity does not have time to develop.
Assumed maximum head rise on valve closure ja=gVj is thus kept
within acceptable limits. An alternative approach is to influence the
way in which the valve door moves, specifically by slowing door move-
ment over all or possibly during the final stages of movement. Reversed
flow is allowed to develop and the door movement is controlled by some
form of ‘damping’ mechanism to prolong closure time. As the door
slowly closes, reversed flow which has developed is then gradually
reduced, yielding a modest deceleration dV=dt and a corresponding
quiet closure.
21.1 Check valve response

Discussions in Chapter 20 relate largely to free-acting valve behaviour.
System response was represented by the rate of deceleration of flow
jdV=dtj and the corresponding variations of pressure at the check
valve. Equal consideration must be given to the ability of check
valves to close in an acceptable manner at a particular rate of flow
deceleration. Not all valves yield the same response. To illustrate this
point Fig. 21.1 shows measured transient pressure just downstream of
the check valve, for three different free-acting valves placed within
409
15 14.7 bar(g) max

P Case 1: Blakeborough swing check valve
(bar(g)) 10 350 mm NB Fig. No. 3820 – light door
15 Case 2: Amri wafer check valve

400 mm NB Series 2000
P 11.4 bar(g) max
Fig. No. 1H3T2KF – light springs fitted
(bar(g)) 10
15
P Case 3: Amri wafer check valve
(bar(g)) 400 mm NB Series 2000
10 8.24 bar(g) max Fig. No. 1H3T2KF – heavy springs fitted
Pump trip Barepot Pumping Station
0 0.5 1.0 1.5 2.0 2.5

Time (s)
Fig. 21.1. Check valve performance comparisons
the same reflux environment at Barepot pumping station. In each case

the duty pump was tripped under the same flow conditions. Reflux
valves which will close silently can be produced without difficulty.
The same valve cannot always be guaranteed to offer minimum
resistance to flow and the two factors are largely incompatible.
To discover why valves respond in different ways in a particular situa-
tion, consideration has to be given to the valve characteristics which
affect movement of the closure element(s). The following sections
examine a range of valve types.
While interest is predominantly centred on check valve closure, it
should be noted that damage has also occurred when free-acting
valves are opened rapidly, for example after pump start. The door can
be thrown open violently to impact against its seat, with very high
deceleration of the door. Door deceleration gradients of 6000 g have
been recorded and valve door stops have been broken.
The following sections of this chapter describe the main features of a
range of valve types.
410
21.2 Swing check valves

Swing check valves are the most common type of check valve found
in water and effluent pipelines, being suitable for both applications
(Fig. 21.2a). It comprises a single hinged door with the ends of the
shaft usually projecting through the valve body. The buoyant weight
moment provided by the valve door is used to assist closure. From a
time of closure standpoint it suffers from a number of drawbacks, parti-
cularly in larger sizes and quite often additional features are added to
the valve to change its closure performance. As a general rule this
valve should be considered for use in solo pumping duty on low to
medium lift systems and on low lift systems where multi-pump duty is
envisaged. Because of its large door opening angle of 708 or greater,
the valve offers an unrestricted flow passage and is suitable for sewage
as well as cleaner liquids.
Figure 21.3 shows a schematic of the valve.
P During closure the valve
door rotates around its hinge with moments ¼ 0. The moments
involved are as follows:
Inertia moment ¼ I d2 =dt2 , where I is the mass moment of inertia
of the valve door (kg/m2 ) and d2 =dt2 is the angular acceleration
of the door (rad/s2 ).
Closing moment ¼ Ws rm sinðÞ, where Ws is the buoyant weight
of the door (N), rm is distance from the hinge to the door centre
of gravity (m) and is the door angle measured from the vertical
(rad).
Bearing friction moment ¼ rp Tr , where is the coefficient of
friction, rp is radius of the hinge pin (m) and Tr is the resultant
thrust on the hinge comprised of the vector sum of buoyant
weight and hydrodynamic thrust (N).
Ð
Hydrodynamic moment ¼ pr dA, where p is the differential
pressure across the element of door area dA and r is the distance
from the hinge pin to the elementary area. Differential pressure
is assumed proportional to the total tangential velocity u2 or
p ¼ C du2 =ð2gÞ with u ¼ V cosðÞ þ r d=dt. A detailed
discussion of the representation of hydrodynamic moment has
been given by Mualla (1983).
As far as performance is concerned, the basic swing check valve
suffers from a number of disadvantages. Considering the buoyant
weight moment term; the light door does not have a large weight and
411
Cover Air plug
Hinge pin Body
Flow
Door
Seat ring Face ring

(a) Swing check valve
(b) External lever and weight (c) Integral relief valve
(d) External lever and axial spring (e) External torsional spring
(f) Air chamber + lever and weight (g) Cylinder and oil reservoir
Fig. 21.2. Swing check valve and modifications
412
m = hinge friction
rm = distance to centre of gravity
rp
Centre of gravity
+V
Seating angle = b
q dq/dt, d2q/dt 2
a = opening angle
Ws = buoyant weight
Fig. 21.3. Schematic of swing check valve door
so this part of the buoyant weight closure moment is modest. A large

opening angle means that the door has a relatively long way to travel
before reaching its seat, implying a greater time of closure. Seating
angle is often close to the vertical when installed in a horizontal pipeline
— the most common application — so that the lever arm rm sinðÞ
becomes very small as the door approaches its seat, rendering the
closure moment almost negligible at this stage. The valve may often
rely to some extent upon an actual reversed flow to push the door
onto its seat. On closure, subsequent abrupt elimination of the reversed
velocity can produce valve ‘slam’ with noise, vibration, high transient
pressure and possible damage.
These features of the swing check give this type of valve a relatively
poor closure performance, especially in larger sizes. Modifications can
be made to this valve in an effort to improve closure performance
and Fig. 21.2b—g illustrates a range of these.
Two fundamentally different approaches can be adopted when
deciding the way in which a check valve will behave during closure.
The free-acting behaviour of the basic valve can be enhanced to yield
a faster response. In these circumstances an ‘ideal’ free-acting valve
would close just at the moment of flow reversal. In practice there will
be some reflux Vo , however small, when the door meets its seat.
Where this velocity produces an inertial head rise jaVo =gj, which is
413
acceptable then the valve is considered satisfactory. On the other hand

if closure is protracted in relation to the rate of deceleration, a sub-
stantial reversed velocity Vo can develop with a correspondingly large
inertial head rise when the valve shuts. Free-acting valve response
can be improved to give a smaller Vo . Modelling of the pipeline
system can be carried out without detailed consideration of the specific
free-acting valve up until the moment of closure so that the methods
described in Chapter 20 can be employed without modification.
The second option to improve closure conditions is to delay the
valve closure and allow the reversed flow to develop. By controlling
the time of closure the reversed flow can gradually be ‘throttled’ up
until the moment of closure. Thus as the reversed flow is progres-
sively eliminated, the deceleration gradient dV=dt can be made as
low as necessary to avoid unacceptable transient effects on closure.
Response of this ‘damped’ valve has to be modelled as an integral
part of the pipeline network, including the characteristics of the
damping mechanism.
Using the common swing-check valve as an example both of the
above types of valve modification are considered. Alterations shown
in Fig. 21.2b—e retain the free-acting nature of the valve while Fig.
21.2f and 21.2g illustrate valve damping mechanisms.
21.2.1 Free-acting modifications

An external lever and weight can be added with the lever at an angle to
the valve door (Fig. 21.2b). This enhances the closure moment and
produces an important closing moment even when the door is close
to its seat. The angle between the valve door and the lever can be
used to alter closure performance or indeed to make it easier for the
valve to open if required, although this is liable to have a detrimental
affect on closure performance. The size of weight and the length of
lever arm also influences performance. The greater the additional
closing moment provided by the lever and weight the shorter the
closure time but with the penalty of increased flow velocity to get the
valve fully opened.
Two forms of external spring can be introduced. A linear spring
connected to the lever (Fig. 21.2d) or a torsional spring on the
projecting hinge shaft (Fig. 21.2e). When the valve is open the spring
is in tension and so additional closure moment is achieved without
significantly increasing the moment of inertia of moving parts. Spring
stiffness will improve performance, with a stiffer spring improving
414
closure performance but with the requirement for a higher velocity to

open the valve against the increased resistance.
A rule of thumb sometimes used is to consider use of measures to
improve closure performance when velocity through the valve exceeds
2.5 m/s.
A pressure relief valve which opens to alleviate transient pressure rise
on valve closure, can be included (Fig. 21.2c). A variation is to have a
bypass around the valve and to include a slow-closing valve within the
small bypass line. The bypass valve remains open until flow reverses,
acting as a pressure relief mechanism. The valve then closes slowly,
gradually decelerating flow in the bypass.
The free-acting swing check valve is best operated with the valve
fully open, as a partly open door may tend to ‘flutter’ causing accelerated
wear of the hinge.
To maintain pump prime a swing check valve or other pattern can be
included at the intake from a wet well. The valve is usually integral with
a strainer in these circumstances. Head loss through the strainer may
be 0:8V2 =ð2gÞ.
21.2.2 Valve damping modifications

An air chamber can be attached to an external lever as shown in
Fig. 21.2f. As the valve opens, air can enter the chamber freely from
the atmosphere. When the valve starts to close, the air in the chamber
is compressed and pressure rises. A resistance force is produced which
increases as the valve door moves towards its seat. Controlled release
of air to atmosphere can be achieved using a flow-regulating valve.
Flow through the check valve is allowed to reverse and later as the
valve door nears its seat this reversed flow is gradually reduced to
zero, giving a quiet closure.
Figure 21.2g shows an oil-filled piston and oil reservoir. When the
valve is opening, a free flow of oil from the reservoir to the piston
cylinder takes place. As the valve closes, a flow-regulating valve
controls the rate at which oil returns to the reservoir. The piston move-
ment is also controlled and with it movement of the valve door. During
the controlled closure a reversed velocity is allowed to develop through
the check valve. As the valve door approaches its seat, this reversed
flow is throttled at a controlled rate with modest deceleration rates
dV=dt to give quiet valve closure.
A modification to the oil cylinder approach is to have the piston only
make contact with the lever during the final stages of closure so that the
415
door moves freely towards its seat until the final stages of closure. The
piston can also be mounted internally to make contact with the door
itself.
A relatively slow-closing valve may allow high reversed velocity to
occur and the damping device may be unable to control the hydrody-
namic forces and can become unstable if not carefully chosen. A
quick-closing free-acting valve may be considered preferable in some
circumstances.
21.3 ‘Recoil’ valves

The closure performance of free-acting check valves having a rotating
door can be improved substantially if a number of changes are made
to the geometry of the valve. These improvements produced the
‘recoil’ check valve, an example of which is shown in Fig. 21.4a. This
valve remains suitable for both clean water and also sewage applica-
tions. The enhancements include:
(a) a weighted door which increases buoyant weight and closure

moment substantially
(b) a restricted ‘lift’ so that the valve door does not open to the same
degree as the simple swing check; this reduces travel distance
(c) a seating angle which is 30—408 from the vertical. This not only
reduces the travel distance of the valve door but also ensures
that a good lever arm and thus closure moment remains even as
a door nears its seat.
In larger sizes the good closure performance of a single-door valve,

which deteriorates as diameter increases, can be retained by having
several doors. Figure 21.4b shows a three-door pattern. Doors of
these larger valves can also be weighted as well as having a similar
seating angle as the smaller sizes. The lift of the valve door can likewise
be restricted. It should be noted that closure performance of these
valves has been optimised for a particular installation and the same
closure performance cannot be expected if the valve is placed in a
vertical riser. However, valves have been produced which are optimised
for vertically upwards flow and Fig. 21.4c depicts such a three-door
type.
These valves are suitable for application to water and sewage
systems with branch velocities typically 3—5 m/s. Recoil types should
be considered for high-lift, multi-pump applications.
416
Diaphragm Stopper
Air plug
Hinge pin Cover
Outlet body
Hinge pin
Inlet body Outlet body
Door
Flow
Flow
Door
Seat
Seat ring
Face ring
Inlet body Face
(a) Glenfield recoil check valve (b) Multi-door check valve
(c) Multi-door check valve for vertically upwards flow
Fig. 21.4. ‘Recoil’ and multi-door check valves
21.4 Tilting disk valve

Also known as a ‘slanting’ disk, this free-acting valve also relies on rota-
tion of the valve door to achieve closure but with the hinge partway
along the valve body (Fig. 21.5).
This valve has a good closure performance on account of its seating
angle and restricted opening angle. Some versions such as the Glenfield
type were specifically designed to have a low head drop.
The valve is available in wafer patterns such as the NAF check so
that reasonable space saving is achieved. The disk is light in weight
with low inertia and opens fully with a velocity of 0.6 m/s, thus avoiding
disk flutter. Seating angle is 158 from the vertical (in a horizontal line)
417
Cover Air plug
Body half Body half
Face ring Disc
Flow
Seat ring
(a) Glenfield tilting disk valve
Stop
Open
Valve fitted with auxiliary

torsional springs
+
(c) Auxiliary torsional spring

Shut
Opening angle
(b) Wafer-type body tilting disk
Fig. 21.5. Tilting disk valves
up to DN 450 and 108 for DN 500 or more. Auxiliary internal springs

can be fitted to enhance closure performance, with a reduction in
pressure rise of 75% after spring addition being measured in one test.
The valve is best suited to clean water applications rather than raw
sewage.
Tilting disk valves are not generally suitable for vertically downward
flow.
21.5 Rubber flap valve

Bearing a superficial resemblance to the swing check type is the
rubber flap valve (Fig. 21.6). This free-acting valve is simple in design
and is suitable for both clean water and sewage applications. The
valve has a door which is encased in rubber and the entire valve
418
Closing disk
Ductile iron/nitrile coated
+
Normal flow direction
(a) Flexible hinge check valve
Flexible hinge

Free closure
+
Damped closure
Optional bottom buffer

(b) Rubber flapper check valve
Fig. 21.6. Rubber flapper check valves
body is sometimes designed to accept a rubber lining. Instead of

having a hinge and shaft, a rubber strip is included. One end of the
strip is connected to the door and the other end is clamped in place
at the top of the valve body. Like the swing check, the door rotates
to open with the rubber strip bending. If the door is light the buoyant
weight closing moment will be more modest than with a heavier
door. The example shown in Fig. 21.6a will not have a good closure
performance and its merits lie in its simplicity. If the seat angle is
about 458 from the vertical, as illustrated in Fig. 21.6b, and angle of
travel is 358 then the buoyant weight moment remains significant
until closure, providing an improvement in closure performance. A
seat angled away from the vertical also helps to reduce overall valve
419
door travel and thus closure time. Since no shaft projects through the
valve body, the options of having external lever and weight or springs
to assist closure are not available with these valves.
In some valves, such as shown in Fig. 21.6b, the option of installing a
‘bottom buffer’ is available. This is in the form of a piston which projects
into the valve near to the bottom of the seat. When the closing door
makes contact with the piston over the final stages of closure,
the piston gradually moves back into an oil-filled cylinder, effectively
throttling any reversed flow which develops and controlling decelera-
tion rate dV=dt. This modification produces a damped closure over
the final stages of movement and it is necessary to know characteristics
of the oil cylinder and piston for any modelling exercise involving valve
closure.
Being rubber-coated, door closure will not be accompanied by the
sharp crack which is present with metal-to-metal seating.
21.6 Split disk valve

Another pattern of free-acting valve which has become popular is the
split disk type (Fig. 21.7). It is composed of two lightweight semicircular
hinged doors. Valve closure is accomplished using a pair of torsional
springs. When installed in a horizontal line with the valve spindle
vertical, there is no buoyant weight moment to assist closure but
reliance is entirely on the springs. When installed in a vertically
upwards flow line, buoyant weight moment of the doors assists closure.
Conversely, in a vertically downwards flow, door weight retards closure.
Due to reliance on internal springs, the use of this type of valve is more
restricted, with installation in sewage schemes not advisable.
Shut Fully open

Torsional springs
Fig. 21.7. Split disk valve with torsional springs
420
In terms of closure performance, the split disk valve is an improve-

ment on the basic swing check but does not match the recoil pattern.
Popularity of the valve rests on its lightweight, wafer design which
can be fitted between flanges, and its relatively lower cost. The space
saving achieved by using this valve can have an influence upon the
dimensions of pumping station buildings.
21.7 Butterfly valve used as a check valve

The butterfly valve can be fitted with an external lever and weight plus
an oil-filled cylinder to damp valve motion (Fig. 21.8). A locking device
is included which prevents the valve closing until pumping failure has
occurred. Typically a butterfly valve might be used on a larger diameter
of line, say DN 1400 or larger. Use of the damping mechanism, as for all
valves thus fitted, allows a substantial reversed flow to occur for an
appreciable time. Reversed rotation of pumps should be investigated.
In one instance the pump was running backwards at 150% of its
design forward speed and it is important to ensure that this is
Door closed
90∞
Flow direction Door open
Oil flow regulating valve
Lever and weight

Oil reservoir
Cylinder connected to lever
Fig. 21.8. Butterfly valve as a check valve (simplified)
421
acceptable. It is possible for the damping mechanism to have a two-

stage closure. For instance a relatively rapid closure may be adopted
between 100% open to 30% open in say 5 s and the remaining closure
from 30% open to shut taking 10 s. Reopening of the valve requires a
hydraulic actuator.
21.8 Nozzle valves

A second group of check valves involves axial movement of the valve
door rather than the rotational movement discussed hitherto. Three
examples are shown in Fig. 21.9. Many of these valves fall into the
category of nozzle valves. An axial spring mounted behind the door is
compressed by the forward flow of water manifest as a differential hydro-
dynamic pressure force across the valve door. When flow decelerates
the hydrodynamic force diminishes, allowing the spring to expand
Linear spring Shut
+ +
Open
Spring Open Shut
(a) Demag type DRV-Z nozzle check valve (b) Nozzle check valve
e.g. SOCLA type 402
Spring
Open Shut
(c) Alstom Clasar check valve
Fig. 21.9. Valves with axial springs
422
and pushing the door on to its seat. The distance which the valve door
has to travel is generally small, allowing a relatively prompt closure to be
achieved.
Closure performance is governed by the equation:
X
axial forces ¼ 0
These pertinent forces are:
ð
Hydrodynamic force ¼ p dA
Friction force ¼ R
Inertial force ¼ Ws =g dv=dt
Buoyant weight ¼ Ws (þ sign in upwards flow and sign in down-
wards flow)
Spring force ¼ x (where is in N/mm and x is movement in mm)
Good closure performance can be achieved with a short stroke and a
small disk mass. An adjustable spring force gives flexibility in applica-
tion of the valve.
As with other internal spring-assisted valves it is deemed inadvisable
to use this type of valve when the liquid carried contains material which
may become entangled with the spring and impede closure. Some of
these valves, such as the Demag nozzle type (Fig. 21.9a), have a stream-
lined body designed to maximise recovery of kinetic energy developed
as flow passes through the throat of the valve. This streamlining
tends to produce a longer valve than other types which do not recover
the same amount of energy. There has to be a trade-off between energy
savings in the case of the more expensive longer-body type for which
increased space must be reserved within a pumping station and a
cheaper valve in which greater head loss occurs but which is shorter
and does not require the same amount of space. The globe-shaped
version is an example of a shorter body type (Fig. 21.9b), with a door
travel of about DN/4.
The Alstom Clasar wafer type illustrated in Fig. 21.9c, is a variation
on the axial movement valve. The valve door comprises a set of
concentric streamlined rings having a central shaft and axial spring.
When closed, these moving rings seal against a set of fixed rings. On
opening, the spring is pulled into tension by the hydrodynamic force
acting on the set of moving rings. Travel of the moving rings is typically
<DN/10 and valve length is <DN/2. When a pump is tripped and flow
423
Spring
Cover
Guide bush
+ Flow direction
Body
Disc Seat
Fig. 21.10. Lift-disk check valve
decelerates, the hydrodynamic force decreases allowing the spring to

pull the moving rings onto their seat. This valve has many advantages
including having a very good closure performance. Being a wafer-type
valve it is short, thus taking up minimal space. As with other valves,
having relatively constricted flow passages and internal springs, the
valve is more suited to relatively clean liquids rather than sewage appli-
cations. Head drop through the valve when fully open is 2:2V2 =ð2gÞ.
One other example of a valve employing a linear spring for closure is
the lift-disk shown in Fig. 21.10. The more complicated flow passage
through the valve tends to yield a greater head loss than other patterns.
A very good closure performance is achieved, making this valve suitable
for applications such as in high-rise buildings.
There remain other patterns of check valve in which the closure
element is more unusual.
21.9 Moving ball

A further example of a free-acting type is the moving ball pattern in
which a spherical door is pushed along guides by flow passing through
the valve, thus moving the door from its seat (Fig. 21.11). When
flow ceases, the ball rolls back down the guides to seal the valve.
This valve is suited to sewage applications but does not have a good
closure performance. Its use should preferably be restricted to systems
424
Valve open
Normal direction of flow Valve closed
Fig. 21.11. Moving ball check valve
where flow deceleration is relatively low at the valve, for instance solo
duty pump installations with modest static lift, so that the system-
imposed dV=dt is small. Advantages are that the ball is self-cleaning
with virtually no risk of blockage. This valve pattern finds application
in sewage systems and other areas such as mine drainage.
21.10 Sleeve or duckbill valve

Many of the free-acting valves considered, such as the swing check
valve, develop the necessary force to decelerate a reversed flow by
mobilising the resistance offered by the valve and its surrounding
pipework and supports. Since these are generally relatively rigid, the
deceleration force is usually developed with only a small elastic defor-
mation of the valve and adjoining pipes. The reversed velocity is
reduced to zero in a very short time after the door meets its seat. The
sleeve valve represents a departure from this type of behaviour. Suitable
for both horizontal and vertical upward flow, this valve contains no
obvious damping mechanism. A rubber sleeve is reinforced either
internally using fabric Nylon ply or possibly externally with longitudinal
stiff rubber strips. These additions act in the same manner as springs. At
design flow the sleeve is opened to give a circular section and as flow
decreases then these stiffening springs pull the rubber sleeve back to
a closed position (Fig. 21.12). The force required to resist and to
decelerate a reversed velocity is achieved by the subsequent deforma-
tion of the sleeve following closure. Up until the moment of initial
425
Cast iron or aluminium body
Neoprene, pure rubber, hypalon, EPT or Buna-N

sleeve reinforced with nylon fabric
Closed Part open Fully open
Direction of
reversed flow
(a) Valve closed (b) Flow reversed (c) Sleeve inverted
Fig. 21.12. Sleeve or duckbill valve
closure the valve acts much as a free-acting valve in that it does not
greatly affect the rate of flow change. Subsequent to initial closure,
deformation of the sleeve over a significant time produces a more
gradual flow deceleration than say a swing-check valve, and this part
of the process is more akin to that of a damped valve.
This valve is suitable for lines conveying slurry, sludge, lime, chemical
slurries, paper plant waste and raw sewage, with the flexibility of the
rubber sleeve allowing the valve to seal even in the presence of
gravel. When installed in appropriate conditions, the valve will produce
a quiet closure. If subject to excessive transient differential pressure
across the sleeve, the possibility of damage exists through the rubber
sleeve being pushed inside out or ‘inverted’. To prevent damage it is
essential to choose a valve having the ability to resist the maximum
differential pressure likely to be experienced across the valve. A range
426
of reinforcement options is available, from very soft where inversion will

occur at differential heads 6 mWG, to much stiffer forms with
maximum allowable differential pressure head 100 mWG. As far as
predicting behaviour of this type of valve is concerned, although the
valve is free-acting in the sense that it does not have any damping
mechanism, nonetheless the response of the valve during closure has
an influence upon the hydraulic transient event and thus upon the
maximum heads developed across the valve. The valve should ideally
be modelled as a part of the pipeline system.
The following represents an attempt to model the hydraulic transient
behaviour of a sleeve valve within a pipeline system.
With reference to Fig. 21.13, after the valve has initially closed, as
differential pressure changes so the deformed shape of the sleeve will
Piezometric line
Hd/s
Valve body
Hu/s
Reversed flow
V = velocity
+
Inverted sleeve Pipe area = A

Inlet volume = Vol in Check sleeve
Vol in
dVol in/dDH = rate of change of inlet volume

with differential head across valve
DH = Hd/s – Hu/s
Fig. 21.13. Definition sketch for sleeve valve analysis
427
alter and with it the volume of liquid on the inlet and outlet sides of the
sleeve. The total volume within the valve body remains constant — that
is, inlet volume þ outlet volume ¼ constant. It is necessary to know the
relationship between inlet or outlet volume and differential head. This
information can be obtained from, for example, experimental measure-
ment or finite-element analysis of the sleeve under load. Given the
relationship between inlet volume and differential head and ignoring
inertial effects within the sleeve, then over a time increment of the
analysis:
VA ¼ dVolin =dt ¼ dVolin =d H d H=dt ð21:1Þ
where Volin is the inlet volume, H is the differential head ¼
Hd=s Hu=s , V is the velocity and A is area of flow. Digitising the
relationship between inlet volume and differential head (Fig. 21.13),
and interpolating using initial values and averaging over the time
increment t then with dVolin =d H Volin =ðHÞ ¼ gradient of
Volin =H curve and suffix ‘o’ denoting conditions at the start of the
time increment:
ðV þ Vo ÞA=2 ¼ gradient ½Hd=s Hu=s ðHd=so Hu=so Þ=t
From the characteristic paths:
Jþ ¼ V þ g=aHu=s and J ¼ V g=aHd=s
or
Hd=s Hu=s ¼ a=g½2V ð Jþ þ JÞ
Writing, constant ¼ 2a=g gradient=ðA tÞ and rearranging then:
V ¼ constant½ Jþ þ J þ g=aðHd=so Hu=so Þ=ð2 constant 1Þ Vo
ð21:2Þ
Corresponding values of Hu=s and Hd=s are obtained by substituting the
value of V into the quasi-invariant equations.
Prior to initial sleeve closure and subsequent to any sleeve inversion,
the relationship between differential head and velocity should be
expressed in the form:
H ¼ KL V 2 =ð2gÞ
where KL is a loss coefficient. The sleeve may become inverted if the
differential pressure exceeds a maximum permitted limit. Transient
analysis can be used to establish this limit for any system so that an
appropriate sleeve can be used.
428
Red valve hmax = 20

3
Series 1
2.5 Series 2
1.5
Velocity (m/s)
0.5
0
0.001
0.121
0.240
0.359
0.478
0.597
0.716
0.835
0.954
1.073
1.192
1.311
1.430
1.549
1.668
1.788
1.907
2.026
2.145
2.264
2.383
2.502
2.621
2.740
2.859
2.978
3.097
3.216
3.336
3.455
3.574
–0.5
–1
–1.5
Time (s)
Fig. 21.14. Velocity at a sleeve valve after pump trip
A flexible sleeve offering only a modest resistance will deform to a

greater extent under the influence of a reversed flow and differential
pressure than will a sleeve of much higher stiffness. Two systems are
considered to illustrate behaviour of the valve.
The first example concerns pumping failure within a short pipeline of
length 24 m and with a static head of 14 m. Pipe diameter was 400 mm.
A number of sleeve stiffnesses were considered, with the stiffness
defined by the maximum allowable differential pressure head which if
exceeded would invert the sleeve.
Enclosed volume within the valve upstream of the sleeve varied
substantially for two stiffnesses represented by maximum allowable
differential pressure heads of 20 mWG and 50 mWG. For the smaller
stiffness the sleeve volume was reduced to around 15 litres from an
initial 58 litres, while the stiffer sleeve reduced in volume to around
41 litres. For the lower sleeve stiffness of 20 mWG, reversed velocity
reaches almost 1.4 m/s after about 1 s as shown in Fig. 21.14. For
the stiffer sleeve, maximum reversed velocity was about 0.8 m/s
after 0.88 s.
Concerning head variations, Fig. 21.15 shows head upstream of the
pump and at inlet and outlet of the sleeve valve for the sleeve with
maximum allowable head of 20 m. Since outlet head is largely deter-
mined by a constant downstream head for this short pipeline system
there is little variation in peak head. At the valve inlet, minimum
head is lower for the 20 m sleeve than for the 50 m head sleeve, because
429
Red valve hmax = 20

16
14
12
10
Head (mAD)
6 Series 1
Series 2
4
Series 3
0
0.001
0.119
0.237
0.354
0.472
0.589
0.707
0.825
0.942
1.060
1.177
1.295
1.413
1.530
1.648
1.765
1.883
2.001
2.118
2.236
2.353
2.471
2.589
2.706
2.824
2.942
3.059
3.177
3.294
3.412
3.530
3.647
–2
Time (s)
Fig. 21.15. Head upstream and downstream of a sleeve valve after pump trip
of the higher rate of deceleration for the much larger reversed velocity
achieved in the 20 m sleeve.
If the reversed velocity is too great, a sleeve may be deformed to an
extent that it becomes inverted. Under these conditions the valve may
no longer remain fully closed and could allow some reversed flow to
continue. The rate of this continued reversed flow will depend upon
the inverted shape and stiffness of the sleeve. To illustrate conditions,
a simple system was considered. A uniform pipeline of length 980 m was
chosen which rises continuously from invert level þ0.0 mAD down-
stream of a pumping station to an elevation þ30.0 mAD at chainage
980 m. Discharge level was set at þ40 mAD. Pipe diameter was
400 mm. Suction well level was 4.0 mAD.
Pump start produced an eventual steady velocity of 1.9 m/s. Alterna-
tive sleeve stiffnesses were examined, ranging from 60 m to 100 m.
After pump trip, velocity at the pump and in the pipeline decelerates
and reverses after about 4 s as shown in Fig. 21.16 for a 60 m sleeve. An
initial reversed velocity is developed, reaching around 0.3 m/s. As the
sleeve deforms in an endeavour to mobilise sufficient strength to
decelerate the reversed flow, it becomes inverted after around 6 s as
shown in Fig. 21.17.
Figure 21.18 shows the head variations predicted for this event. After
trip, head downstream of the pump falls to 10.0 mAD and thereafter
increases gradually. After around 4 s the sleeve becomes closed and flow
430
Red valve – longer system hmax = 60

2.5
Series 1
2 Series 2
1.5
Velocity (m/s)
0.5
0
0.005
0.353
0.701
1.049
1.396
1.744
2.092
2.440
2.788
3.136
3.484
3.832
4.180
4.526
4.875
5.223
5.571
5.919
6.267
6.615
6.963
7.311
7.659
8.007
8.354
8.702
9.050
9.398
9.746
10.094
10.442
–0.5
–1
Time (s)
Fig. 21.16. Velocity at failing sleeve valve
reverses, with head at the valve outlet rising steadily to a maximum

when the differential head across the valve reaches 60 m, at which
point the sleeve is considered to invert quite rapidly. Depending
upon the final shape of the inverted sleeve, reversed flow will continue
to proceed through the valve, producing a fall in valve outlet head.
Thereafter, conditions gradually approach a final steady reversed flow
condition within the system as a whole. The continuous reversed
Red valve – longer system: cv = 0.08, hmax = 60.0

60
Series 1
50
40
30
Volume (l)
20
10
0
4.091
4.170
4.248
4.327
4.405
4.483
4.562
4.640
4.719
4.797
4.875
4.954
5.032
5.111
5.189
5.267
5.346
5.424
5.503
5.581
5.659
5.738
5.816
5.895
5.973
6.051
6.130
6.208
6.287
6.365
6.443
6.522
–10
–20
–30
Time (s)
Fig. 21.17. Downstream sleeve volume (computed)
431

60
50
40
30
Head (mAD)
20 Series 1
Series 2
10 Series 3
Series 4
0
0.005
0.475
0.946
1.416
1.886
2.357
2.027
3.218
3.768
4.218
4.749
5.109
5.650
6.120
6.590
7.061
7.581
8.002
8.472
8.942
9.413
9.883
10.354
10.824
11.294
11.765
12.285
12.706
13.176
13.646
14.117
14.587
–10
–20
Time (s)
Fig. 21.18. Head at failing sleeve valve after pump trip
flow has the consequence of producing reversed pump rotation as

illustrated in Fig. 21.19.
A stiffer valve sleeve is required to avoid inversion. Increasing the
maximum allowable differential head to in excess of þ80 m to say
100 m produces the desired effect. Figure 21.20 shows velocity changes
at the valve and at chainage 500 m. An oscillation in velocity was
predicted to develop following sleeve closure after about 4.7 s, with

1.2
Series 1
1.0
0.8
Pump speed/design speed
0.6
0.4
0.2
0.0
0.005
0.480
0.955
1.431
1.906
2.381
2.857
3.332
3.807
4.283
4.758
5.233
5.708
6.184
6.659
7.134
7.610
8.085
8.560
9.036
9.511
9.986
10.461
10.937
11.412
11.887
12.363
12.838
13.313
13.789
14.264
–0.2
–0.4
–0.6
–0.8
–1.0
Time (s)
Fig. 21.19. Pump speed for sleeve valve failure
432

2
Series 1
Series 2
1.5
Velocity (m/s)
0.5
0
0.005
0.603
1.200
1.798
2.396
2.994
3.592
4.189
4.787
5.385
5.983
6.581
7.178
7.776
8.374
8.972
9.570
10.167
10.765
11.363
11.961
12.559
13.156
13.754
14.352
14.950
15.548
16.145
16.743
17.341
17.939
18.537
–0.5
Time (s)
Fig. 21.20. Velocity variation at sleeve valve in longer pipeline
this oscillation taking place around the zero flow condition. A similar
velocity variation occurs at the valve inlet and with no continuous
reversed flow there is no motivation for reversed pump rotation to
occur. The predicted oscillation was due to substantial deformation of
the sleeve. Under the action of transient effects within the pipeline,
only a 10 litre margin remains before inversion could occur. Head
changes are shown in Fig. 21.21, with peak head of þ75 m at the

80
Series 1
70 Series 2
Series 3
60
Series 4
50
Head (mAD)
40
30
20
10
0
0.005
0.612
1.220
1.818
2.430
3.043
3.650
4.258
4.866
5.473
6.081
6.688
7.296
7.904
8.511
9.119
9.726
10.334
10.942
11.549
12.157
12.764
13.372
13.080
14.587
15.195
15.802
16.410
17.018
17.625
18.233
18.840
–10
Time (s)
Fig. 21.21. Head variations at sleeve valve in longer pipeline
433
valve outlet. A large oscillation was predicted following deceleration of

the reversed velocity.
It should be noted that use of a sleeve which allows inversion to occur
may produce lower system pressures generally than when a stiffer sleeve
is used which prevents inversion. However, the inverted sleeve is liable
to suffer damage and there is the matter of inverse pump rotation to
consider.
21.11 Membrane valve

The closing system for this type of valve is in the form of a flexible
membrane (Fig. 21.22). During flow the membrane deforms and is
held at its centre on a perforated seat. On flow reversal, the mem-
brane is held against a fixed diaphragm, usually of coated steel. The
diaphragm has openings of area approximating the pipe area. The
thickness and elasticity of the membrane can allow progressive open-
ing and closing, rendering this valve suitable for pulsating operation
or in systems having variable flow pumps. Closure is not dependent
upon gravity and so the valve can be used in all positions. The
valve may be considered for use in circumstances where deceleration
is rapid.
Perforated steel seat
Steel coated grill Flexible membrane
Fig. 21.22. Membrane valve
434
21.12 Prediction of valve behaviour

The variety of valve designs makes it impossible to have a standard
approach to assessment of valve behaviour. Damped valves and the
sleeve or duckbill valve can be studied by including the valve as part
of the pipeline system into which the valve is intended to be installed.
However, other valves of the free-acting type are amenable to having
their closure performance presented in the form of charts which are
independent of any specific pipeline system. This section describes
closure behaviour for a range of valve types. Many valves remain
unknown in terms of their dynamic performance and even a relatively
modest change in a documented valve’s configuration can produce a
significant change in its performance.
The potential for many patterns of check valve to produce shock
pressures following a delay in valve closure has been long recognised
and extensive studies have been undertaken to improve valve design.
As early as 1954, Livingston described studies undertaken by Glenfield
and Kennedy which included measurements of the hydraulic moment
acting on swing check and tilting disk valve patterns. These studies
led to the development of the ‘recoil’ valve. The year 1959 saw con-
siderable activity in studying check valve performance with Whiteman
and Pearsal measuring closure behaviour of swing check valves at
Kingston Power Station; Worster (1959) and also Esleek and Rosser
(1959) proposed means of calculating dynamic response of a reflux
valve door in a decelerating and reversing water column. In 1962
Pool et al. used a modification of the work of Esleek and Rosser and
produced good agreement between prediction and observation of
valve closure motion. Confidence in their model was sufficiently high
for it to be used in the analysis of check valve performance in both
thermal and nuclear power plants. During his 1965 studies, Parmley
used the techniques developed by Pool et al. to predict response of
swing-check valves, Recoil valves and tilting disk valves under reflux
conditions. His predictions showed good agreement with observations
obtained from a transient test facility constructed by Glenfield and
Kennedy. Despite these studies, problems associated with check
valves continued to occur and in his opening address to the second
Conference on Pressure Surges in 1976, Lupton (1976) appealed for
further research and development into check valves and suggested
that up to 50% of waterhammer incidents could be attributed to this
source. Since his address, numerous studies, both experimental and
numerical, have been undertaken and dynamic closure performance
of a large number of valve patterns has now been quantified. Despite
435
the research into valve behaviour, incidents of unacceptable valve

performance still continue to occur.
Model testing plays an essential role in quantifying dynamic perfor-
mance of check valves and around the world test facilities have been
constructed which allow a check valve to be subject to varying decelera-
tion rates in order to establish how that valve will respond over a range
of operational circumstances. An early example was that of Glenfield
and Kennedy and among later installations were those of CEGB at
Marchwood in England and Delft Hydraulics. Delft laboratory has
been especially active in examining different valve patterns and in
working to establish a means of presentation for test results. By testing
different patterns of valve in this way, the relative merits of each valve
can be gauged and appropriate limits of operation, in terms of allowable
gradient jdV=dtj, defined. In the absence of information on dynamic
performance, check valves have often been selected on the basis of
price and availability. However, increasingly engineers are specifying
the need for ‘non-slam’ performance. ‘Slam’ refers to the noise gener-
ated on closure and is a function of the dynamic response of the
valve. However, valves come in a range of sizes. Take, for example,
the popular ‘split-disk’ type. This may be obtained in diameters ranging
from 40 mm to 1800 mm. Dynamic closure performance deteriorates
with increasing diameter and so a means of extrapolating findings
from tests on one valve size to a geometrically similar family of valve
diameters is required. Further complications may also arise, for instance
from a range of spring stiffnesses, or different properties of lever, weight
and lever/door angle, all of which can significantly alter the dynamic
response of a valve. Mathematical modelling has a role to play by
allowing a diameter of valve which has been tested to be treated as a
model for other valve sizes.
Before any reliance can be placed upon computational models, the
predictions must be compared with observations. Figure 21.23 shows
some typical comparisons between model predictions and laboratory
measurements for Glenfield Valves of the swing check, tilting disk
and recoil patterns. These valves were all installed in a horizontal pipe-
line. In this figure, H is the pressure head just downstream of the check
valve, is the valve door angle measured from the vertical and V is the
velocity of flow in the pipe at the check valve. The subscript ‘m’ denotes
measured valves and subscript ‘c’ indicates a computed value.
A convenient means of presentation for many but not all valve types,
is to plot deceleration gradient jdV=dtj against the reversed velocity Vr
at the time of valve closure. Results both directly from physical testing
436
H (m) q (deg) V (m/s)
60 60 qm qc 6
50 50 Swing check DN 225 5
40 40 4
30 30 3
Vm Vc
20 20 Hm 2
Hc
10 10 1
0 0 0
0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6 1.8 2.0
70 70 7
qm qc
60 60 6
50 50 Tilting disk DN 225 5
40 40 4
Vm Vc
30 30 3
20 20 Hm 2
Hc
10 10 1
0 0 0
0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6 1.8 2.0

60 60 6
qm qc
50 50 5
40 40 Recoil valve DN 225 4
30 30 3
Vm Vc
20 20 Hm Hc 2
10 10 1
0 0 0
0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6 1.8 2.0
Fig. 21.23. Measured and computed behaviour at closing check valves
and supplemented by computational predictions have been presented

for different valve types. In any specific pipeline investigation, the
value of deceleration gradient jdV=dtj will be obtained from hydraulic
transient investigation. Entering the curve at the appropriate point
437
along the horizontal jdV=dtj axis using the gradient obtained by

analysis, the intercept with the appropriate curve is obtained and the
value of reversed velocity Vr at closure is found.
Figure 21.24 shows results from studies of swing check valves of
differing patterns. The Keystone Figure 85 is a wafer type with a seating
2.5
400 300
DN 100 Delft hydraulics data
250
Keystone Fig. 85 DN 200 200
Delft hydraulics data
2.0 150
1
Effect of shifting centre of 2
3
gravity of valve door
4
Magnitude of reversed velocity at closure (m/s)
1.5
Glenfield M1
1.0 Glenfield M1 DN 225

Pivot
40∞ Weighted lever

Fully open door position
44∞
Seat angle 16.1∞

Vertical
0.5 1 Initial velocity 1.6 m/s no wt.

2 Initial velocity 2.7 m/s 44.53 N
Lever length 508 mm
0.0
0 5 10 15
Deceleration gradient |dV/dt| (m/s2)
Fig. 21.24. Performance curves for swing-check valves
438
angle close to the vertical when installed in a horizontal pipeline. By

contrast, the Glenfield M1 has a seating angle 16.18 from the vertical
in a similar installation and a 448 angle of turn to the full open position.
A spread in performance results from differences in size, opening angle
and seating angle. Valve performance deteriorates as opening angle
increases. Door weight and other factors such as a lever and weight
may also influence closure performance. When a lever and weight are
added (curves 2—4), the performance is improved for lower deceleration
rates up to around 6 m/s2 but for steeper rates of retardation the effect of
additional inertia caused by the weighted lever produces deterioration
in performance. In one of many tests Provoost (1980) moved the
centre of gravity of the check valve towards the pivot and measured
a deterioration in performance as shown in the figure. Application of
swing check valves will usually be restricted to modest deceleration
gradients unless special measures are introduced such as damping of
2.0
DN 600 recoil type
1.5
Glenfield recoil
series 5100
DN 300
1.0
Glenfield tilting disk

DN 225 DN 250
DN 200
0.5
DN 150
DN 100
0
0 5 10 15
Fig. 21.25. Performance curves for Glenfield recoil and tilting disk valves
439
valve movement. A maximum reversed velocity <1.0 m/s would seem

reasonable.
The recoil type of valve represents a derivation of the swing check,
with a considerable improvement in performance (Fig. 21.25). The
curves shown are a combination of findings from both measurements
and computer simulations. Performance deteriorates with size, possibly
for the same reason as for the swing check with lever and weight.
Deceleration gradients of 10 m/s2 or more are considered acceptable
without risk of valve slam.
Also contained in Fig. 21.25 are limited data for the Glenfield tilting
disk valve DN 225. Performance of this valve is not as good as the recoil
type but has been designed to fulfil a different function.
Measurements from the test facility at Delft Hydraulics on the
moving ball type valve, reported by Provoost (1980, 1983), are shown
in Fig. 21.26. Closure performance is relatively poor, with substantial
2.0
DN 100
from Delft hydraulics data
1.5
DN 200
from Delft hydraulics data
1.0
0.5
0
0 5 10
Fig. 21.26. Performance of moving ball valves
440
reversed velocity at closure attained with deceleration gradients

<5 m/s2 . While other valves show a deterioration in performance
with increasing size, the curves in Fig. 21.26 show an improvement
for the large valve tested. It may be expected that ball weight and
rolling friction play some part in determining closure time, or some
2.4
2.3 DN 450
DN 750
DN 600
2.2
2.1 Duo-chek DN 225

spring stiff = 0.0 Nm/deg
2.0
DN 300
1.9
1.8
1.7
1.6
1.5 Duo-chek DN 225

1.4
1.3 Duo-chek DN 225

1.2
1.1 Duo-chek DN 225

1.0
DN 150
0.9
0.8 Gestra BB16 DN 200

super-strong springs
0.7 Valve open at velocity = 4.8 m/s
0.6 1 broken spring

Duo-chek DN 150
0.5
Duo-chek DN 200 weak springs
0.4
Duo-chek DN 150 vertically upward flow
0.3
Gestra BB16 DN 200 weak spring
0.2
Valve open at velocity = 0.5 m/s
0.1
Duo-chek DN 200, strong springs
0.0
0 2 4 6 8 10 12 14 16 18 20 22 24 26 28 30
2
Deceleration gradient |dV/dt| (m/s )
Fig. 21.27. Performance of split-disk valves
441
other factor such as different ball density may be a significant factor.

The wafer-type split-disk valve with torsional springs to promote
closure has received a considerable amount of attention. The majority
of studies have been conducted for the most common case of a
horizontal valve installation. Figure 21.27 depicts results from different
tests carried out at Delft Hydraulics Laboratory and reported by
2.5
DN 800
DN 700
2.0 Velocity when fully
open = 1.5 m/s
Demag
DRV-B
1.5
DN 600
Measured DN 800
medium spring
DN 500
1.0
DN 400
DN 300
0.5
0.0
0 2 4 6 8 10 12 14 16 18 20 22 24 26 28
Fig. 21.28. Performance of Demag DRV-B valves
442
Provoost (1983), by Collier and Hoerner (1983) and by Mualla (1969).

Also included are computer predictions by Ellis and Mualla (1986) (Fig.
21.27) for different spring stiffnesses and valve sizes. Pake (1983) has
provided results for the DN 150 Duo-chek valve for more unusual
cases of vertically upwards flow and for a valve with one of the torsional
springs removed (broken). The improvement in performance in verti-
cally upward flow reflects the influence of buoyant weight of the
valve doors. Vertically downward flow would be expected to show a
poorer performance than the horizontal installation, as the door
weight moment then acts against closure. The curves of Fig. 21.27
show a clear deterioration of performance with increasing diameter
and as would be expected an improved closure performance as the
spring stiffness is increased.
Another valve type which has received considerable attention is the
Demag nozzle pattern. Several variations of this valve are available and
results for DRV-B and DRVg types are included. Figure 21.28 depicts a
set of computer predictions by Perko (1986) for the DRV-B and also
experimental data for the DN 800 valve from Koetzier et al. (1986).
For the DRV-B DN 300, Fig. 21.29 shows how closure performance
1.3
1.2
1.1
No spring, Vo = 0.0 m/s
1.0
0.9 Demag DRV-B DN 300
0.8
0.7 Normal spring, Vo = 1.5 m/s
0.6
0.5
0.4
0.3
Stiff spring, Vo = 3.0 m/s
0.2
0.1
0.0
0 2 4 6 8 10 12 14 16 18 20 22 24 26 28
Fig. 21.29. Performance of Demag DRV-B DN 300 valve
443
changes with varying spring stiffness. In this figure the initial velocity is
representative of the linear spring stiffness, with a broken or missing
spring requiring zero velocity to maintain the valve in the open position.
A stiff spring by contrast needs a flow velocity of 3.0 m/s to fully open
the valve. Not surprisingly, the stiffer spring is able to close the valve
quickly but the absence of a spring requires differential pressure
1.2 Gestra RK 56
DN 200 weak spring
1.1
1.0
0.9
Demag DRV-B
DN 300 weak spring
0.8
0.7
0.6
0.5
Demag DRVg
DN 200 weak spring
0.4
0.3 Alsthom Clasar

DN 800
0.2
0.1
Demag DRV-B
DN 300 strong spring
0.0
0 2 4 6 8 10 12 14 16 18 20 22 24 26
Fig. 21.30. Performance of different valves with spring-assisted closure
444
across the valve door to achieve movement and closure resulting in a

poor performance.
Figure 21.30 shows results from Delft Hydraulics for a number of
valves having a linear spring closure action. The Gestra RK 56 is of
the globe type as illustrated in Fig. 21.9b. The Clasar type is a wafer
pattern as depicted in Fig. 21.9c. Despite its size (DN 800), the
Clasar valve still retains a very good closure performance.
21.13 Use of the charts

To use the charts requires a preliminary hydraulic transient analysis to
be completed. For the free-acting valve types this can be done without
reference to any specific valve pattern. The primary task of this initial
analysis is to yield a value of the deceleration gradient jdV=dtj at the
proposed check valve site. This value can then be used to enter any
chart at a point on the horizontal axis. Choice of a velocity gradient
may be relatively straightforward in many cases where the coast-down
of velocity follows a relatively familiar pattern. There is an initial
curved transition from steady flow to an almost uniform gradient.
This uniform gradient could be used to give a reasonably conservative
estimate of jdV=dtj. In other circumstances, matters may not be so
clear-cut, with the deceleration of flow at the NRV following a series
of steps for example. In these cases, possibly it would be prudent to
choose a gradient which represented the steeper part of the curve.
Having chosen a value of velocity gradient, then travelling vertically
upwards from this point on the horizontal axis, an intercept on any
curve can be obtained. The curve used will be for the particular size
and type of valve being considered. More than one chart may be
used where several possible valves are under consideration.
From the intercept with the curve a horizontal line can be drawn to
meet the vertical axis from which a value of reversed velocity can be
read off. This velocity can be used in a number of ways.
It may be sufficient to examine the magnitude of this velocity and
decide if it is acceptable.
A calculation of jaV=gj can be made using the reversed velocity to
determine inertial head rise. To this can be added the prevailing
head at the time of valve closure and a total head value established.
The computer model can be rerun with the valve set to close when
velocity reaches the value established from the chart. Peak pressures
and transient behaviour can then be modelled in more detail.
445
22
Flexible pipe
Flexible pipes add another layer of complexity through changes in pipe

geometry over time. Where plastic pipes are concerned the phenom-
enon of ‘creep’ also has a role to play. Deformation of a flexible pipe
causes changes in its ability to resist buckling. Propagation of pressure
waves may also be affected by changes in pipe cross-sectional shape.
Response of flexible pipes laid below ground also requires consideration
of the surrounding fill material properties.
22.1 Review of pipe materials and properties

The equation for conservation of mass commonly used in development of
the quasi-invariant relationships which apply along characteristics,
considers the cross-sectional shape to remain essentially unchanged. It
may expand or contract with varying strain in the pipe wall and the
area within the pipe will vary for this reason. Pipes which may reasonably
be assumed to retain their commonly circular shape usually include those
made of the following materials: grey cast iron, concrete and asbestos
cement. Load redistribution and stress reallocation do not occur to any
great extent in these materials and failure is usually through brittle
fracture. Other pipes may be more flexible by virtue of lower deformation
modulus and/or by having a thinner pipe wall. These more flexible
structures include thin-walled steel pipes, thermosetting materials such
as glass-reinforced plastic (GRP), thermoplastics including polyvinyl
chloride (PVC), polyethylene (PE) and polypropylene (PP), and various
composite pipes involving a combination of plastic materials. For compo-
site pipes such as Permastran which has a PVC core for watertightness
and the glassfibre outer wrap for strength, the glassfibre wrap gives the
composite eight times the strength of the PVC.
446
Flexible pipe
Table 22.1. Allowable pipe strain (Petroff (1984))
Material Allowable strain (%)
FRP 0.3
RPM 0.3
PVC 1.5
HDPE 4.0
Plastic pipes also have the property of ‘creep’ or visco-elasticity, produ-

cing a time-dependent stress—strain relationship while the load is applied.
Some thermoplastics can sustain relatively large deformations without
any difficulty for the material. For example, a DN 900 mm high-density
polyethylene (HDPE) pipe may have an allowable deflection of 20%
with respect to strain; however, stability considerations limit deflection
to a safe value of 7.5%. Long-term diametrical deflection limits of 5—
7.5% can be permitted without problems for joints, soil movement, etc.
Basic consideration for designing an installation of flexible conduits is
control of deflection. For underground pipes a load distribution can be
assumed by mobilising soil support, and concentrated loadings can be
relieved by this process. As an example, HDPE can endure deflections
of up to 30% of diameter without excessive bending stresses. When
buried, such large deflections would not be acceptable due to soil
movement. HDPE elongation can be as much as 500% at failure.
Thermoplastic pipes are characterised by relatively large allowable
strains, high ductility, large deflection without cracking and low stiff-
ness. Typically thermoplastic pipe stiffness will be 317 kN/m2 .
Other plastic pipes of GRP or reinforced plastic matrix (RPM) can
fail due to a lack of ductility. Failure may be through delamination of
pipe walls due to excessive stress. As pipe stiffness is reduced, ductility
must rise to withstand the stresses caused by ovalisation, soft and hard
spots on bedding and other irregularities. Typical allowable strains for
different plastics have been given by Petroff (1984) (Table 22.1).
All flexible pipes display the ability to deform from a purely circular
cross-section under load. Plastic pipes also exhibit ‘creep’ or plastic
deformation. For instance, the shape can change under gravity from a
purely circular form to an elliptical shape merely by virtue of being
stacked while awaiting use. The shape of a deformed cross-section
depends upon the pipe surroundings. Flexible pipes which are above
ground, or in a trench backfilled with only slightly compacted or
dumped bedding or soft clay, or laid under water as in an outfall, will
447
DN 450 (18") steel pipe DN 450 (18") steel pipe

EI/D m3 = 8.618 kN/m2 EI/D m3 = 1.724 kN/m2
elliptically deformed rectangularly deformed
21% maximum diametrical deflection 18% maximum diametrical deflection
(a) (b)
Fig. 22.1. Observed deformed shapes
have a generally elliptical cross-section (Fig. 22.1a), with the ratio of

horizontal/vertical deflection h=v 0:91. When a pipe bedding is
compacted to a moderate or high degree then the initial deformation
may show an increase in vertical diameter by up to 5% as sidefill is
compacted. Backfilling above the pipe then reduces this vertical
elongation in diameter. The resulting shape may either be elliptical as
in Fig. 22.1a or a more ‘rectangular’ form may be adopted (Fig.
22.1b), when pipe stiffness is small compared to sidefill stiffness. The
resultant ratio of horizontal/vertical deflection may be around
h=v 0:2. The influence of the increase in horizontal diameter
extends to about 2.5 diameters so that trench sidewall strength may
also have a significant effect.
22.2 Pressure transient effects

As far as pressure transient behaviour and prediction are concerned,
several significant implications stem from the departure of the cross-
section from its original circular form. Flattening or ‘ovalisation’ of a
circular section will decrease resistance to buckling with an elliptical
or other deformed shape being more prone to collapse than a purely
circular shape in otherwise identical circumstances. If compressive
stress in the pipe wall is too great, failure can occur due to instability.
For underground pipe there is usually a stabilising effect from
surrounding soil, particularly granular soils. Deflection and irregularities
of shape, for example flattening of the pipe crown, reduces stability
considerably. Both long-term buckling and elastic short-term buckling
are possible.
As pressure within a pipe changes with time, the shape will continue
to deform. For example, if pressure falls, an elliptical section will flatten
448
Flexible pipe
further and if internal pressure increases, re-rounding of the section will

tend to occur. Changing shape as a consequence of ovalisation is
accompanied by change of cross-sectional area and a corresponding
variation in pressure wave speed. In the case of thermoplastic pipes,
the phenomenon of ‘creep’ or plastic deformation continues to act in
the long term to produce changes in cross-section shape. The behaviour
of pressure transients will depend upon the cross-sectional form at the
time when the surge occurs. If some reduction in buckling pressure is
considered due to ovalisation while retaining the wave speed for a
purely circular section, the effect may be to predict a lower minimum
transient pressure than will actually occur, resulting in a conservative
design. Ideally, adoption of a reduced resistance to buckling should
be accompanied by the corresponding influence upon wave speed or
damping of the pressure wave.
For non-circular shapes such as square ducts and other polygonal
forms, deformation from the original shape can occur as straight sides
are deformed by bending as described by Thorley and Enever (1979).
The following questions are pertinent for flexible pipelines.
(a) How does the deformed shape influence the propagation of pressure
waves and the Reimann quasi-invariants along characteristics?
(b) How is the ability of the pipe to withstand transient pressures
influenced by the changing shape of the pipe in both short and
long term?
(c) How do repeated pressure fluctuations influence the life expectancy
of a plastic pipe?
(d) How does possible scratching of the outer surface of the pipe
influence pipe strength?
22.3 Strain and deflection

Plastic pipes are probably best analysed by strain rather than stress.
Critical strain for a perfectly round pipe "cr is given by equation
(22.1):
"cr ¼ 1=ð1 2 Þ1=SDR2 ð22:1Þ
the pipe remaining round till this point. SDR ¼ ðoDm =s 1Þ, where oD
is the mean outside pipe diameter.
The critical buckling pressure equation can be written to include
strain. Effective maximum strain "f is given by:
"f ¼ ð1=Dm 1=Dmax Þs
449
or
"f Dm =s ¼ 1 Dm =Dmax
giving,
Dm =Dmax ¼ 1 "f Dm =s ð22:2Þ
where Dm is the original diameter of the round pipe and Dmax is the
maximum deformed diameter.
One of the most difficult tasks is prediction of deformation and strain
over time. A relationship which according to Moser (1981) may be used
for deflections of up to 10% in PVC pipes is:
" ¼ ðs=Dm Þ½3 y=Dm =ð1 2 y=Dm Þ ð22:3Þ
where y is the vertical change in diameter, Dm is mean diameter and "
is the maximum strain. This equation does not include the influence of
internal pressure. If a cylinder is deformed and therefore not perfectly
circular it is inherently weaker and will collapse at a differential pressure
somewhat below that of a circular pipe in the same circumstances.
When a pipe is relatively unconstrained, for example above ground or
when placed in loose soil, the maximum strain may be:
Design value ¼ 0.5%
in emergency situations ¼ 1.5%,
where repeated surging occurs ¼ 1.3%
The ultimate strain is defined as the initial strain at first installation
which would cause failure after 50 years. For PVC this has been
suggested as 1%. Allowable design strain is half the ultimate strain.
Moser could find no basis for this 1% ultimate strain value.
The relationship between pipe ring deformation and tangential strain
in the pipe wall has been expressed in a different form by Molin (1981)
as:
"r ¼ pDm =ð2sEÞ Df ðy=Dm Þðs=Dm Þ ð22:4Þ
where "r is the tangential strain in the pipe wall, y=Dm is the deforma-
tion of the pipe ¼ , s is the wall thickness and Df is the deformation
factor.
This equation represents a linear combination of strains due to
internal pressure pDm =ð2sEÞ and strain due to elliptical deformation
Df ðy=Dm ÞðsDm Þ. If the deformation is elliptical then Df ¼ 3. In
most practical conditions 4 Df 6, although values of Df in some
circumstances may range from 2.85 to 10 or greater. In circumstances
450
Flexible pipe
where pipes have been intentionally poorly installed, Df values of

10—20 have been found.
For design purposes for PE pipes, fos=ð1 2 Þ ¼ 2 has been proposed
(where fos is factor of safety), giving finally:
" ¼ pDm =ð2sEÞ 6=SDR ð22:5Þ
The value ‘6’ is the initial value for a careful installation.
22.4 Establishing the rate of ovalisation in the longer term

Collapse rate R ¼ rate of increase in % age ovality can be expressed as:
log10 ðRÞ ¼ 4:655 2:106 105 p SDR3 ð22:6Þ
where R ¼ minutes/(% age decrease in minimum diameter).
Bishop and Lang (1984) provided the following Table 22.2 of pipe
deflection and shape values (Df and ) as a function of embedment.
For long-term use Df should be multiplied by a factor of 3.
A buried plastic pipe, after reaching the equilibrium state, is held at
constant deflection. For a material such as PVC, which relaxes over
time, once equilibrium is reached internal stress in the pipe wall will
decay with time. Although deflection may be less than 5%, equilibrium
state is only slightly influenced by pipe material properties but is
primarily controlled by initial soil density in the pipe zone.
The vertical diameter of a flexible pipe can increase by up to 5%
due to compaction of bedding soil at the sides of the pipe. When
backfilling on top of the pipe, the vertical diameter reduces. For
elliptical deformation, horizontal deflection is 91% of vertical deflec-
tion. If soil stiffness at the sides of the pipe is large compared to pipe
stiffness, the pipe can deform rectangularly. Horizontal deflection can
be vertical deflection with horizontal deflection 20% of the vertical
deflection. Horizontal deflections > vertical deflections have also been
observed.
Table 22.2. Deflection and shape v. embedment (Bishop and Lang (1984))
Embedment Deformation factor (Df ) ¼ y=Dm
Compact sand 5.30 1.50

Compact native soil 10.30 1.13
Loose sand 4.50 7.04
Loose native soil 4.53 9.24
451
P P
s s Dy
Dy
Dm Dm
(a) (b)
Fig. 22.2. Deformed shapes for alternative side support
For dumped beddings and modest degrees of compaction an elliptical

deformed shape should result with horizontal deformation about 90% of
vertical. This applies to flexible steel pipe as well as plastic pipes.
Deformation up to 5% of the initial outside diameter was considered
acceptable. Assuming simplified values for the supporting effect of fill
and its internal friction for HDPE pipes, the vertical deflection
y m was given by Hoechst Plastics (1971) as:
y ¼ 5 105 ðq=EÞðDm =sÞ3 ð22:7Þ
where q is vertical load on the pipe N/m length and E is modulus of pipe
material N/m2 .
Considering pipes without sidefill support (Fig. 22.2a), the
corresponding deformation can be calculated from:
y ¼ 22:3 105 ðq=EÞðDm =sÞ3 ð22:8Þ
where the pipe was supported by a stiff sidefill (Fig. 22.2b), the deforma-
tion is given by:
y ¼ 3 105 ðq=EÞðDm =sÞ3 ð22:9Þ
The laterally supported pipe undergoes only 13.3% of the deformation
suffered by the unsupported pipe.
22.5 Long-term buckling pressures — unconstrained

surroundings
Considering first the case of a pipeline which is either above ground,
under water or in a trench where the fill is loose or soft. Then the
deformed shape may be taken as elliptical. For ring stiffness/unit
452
Flexible pipe
length of pipe:
Sr ¼ EI=r3 ð22:10Þ
3
where I ¼ s =12.
As regards the allowable buckling pressure within a deformed pipe
Timoshenko (1936) and others have studied the elastic stability of
cylindrical shells and predicted buckling of a round pipe using the
following equation:
pb ¼ 24EI=½ð1 2 ÞoD3 ð22:11Þ
where pb is the critical buckling pressure; E is the modulus of
elasticity of the pipe material; is Poisson’s ratio for the pipe
material, for HDPE ¼ 0:4; I is moment of inertia of the original
circular cross-section ¼ s3 /12; and oD is mean outside pipe diameter.
Where out-of-roundness has occurred, a reduction factor c is applied,
giving critical pressure pc ¼ pb c. Alternative expressions for the reduc-
tion factor c have been developed. These reduction factors may include
corrections both for ellipticity as well as variations in pipe wall thick-
ness. Considering flexible pipes ‘creep’, effects were accounted for by
using the long-term secant modulus and long-term Poisson’s ratio for
pipe material. For a plastic pipe displaying ‘creep’, any calculation of
long-term deflection or buckling strength requires use of the ‘creep’
modulus. For HDPE the creep modulus at 50 years is typically 0.2—
0.3 GN/m2 . When calculating initial deflection and strength for
HDPE, the short-term modulus is 0.8—1.0 GN/m2 . The modulus for
polyethylene may be calculated from equation (22.12):
E ¼ 10:0½5:88 0:067 log10 ðhoursÞ 0:6 log10 ð8FÞ ð22:12Þ
The reduction factor c is given variously by:
c ¼ ðoDmin =oDmax Þ3 ðsmin =smax Þ3 ; that is, c always <1 ð22:13aÞ
Subscripts ‘max’ and ‘min’ denote maximum and minimum values of
diameter and wall thickness.
Ignoring the influence of variations in pipe wall thickness, Allman
(1975) reported an experimentally derived expression for Alydn R ‘D’
PE pipe giving reduction factor ca as:
ca ¼ ðoDmin =oDmax Þ4:62 ð22:13bÞ
Again, neglecting the effect of wall thickness changes, Gaube et al.
(1974) gave a reduction factor cg :
cg ¼ ½ð1 Þ=ð1 þ Þ2 3 ð22:13cÞ
453
where ¼ D=Dm is out-of-roundness. D ¼ oDmax oDm . Values

used are those for the pipe as received or when first installed.
For profile wall pipes then an equivalent solid wall thickness is
required for use in the equations for buckling pressure.
Above-ground mains subject to internal vacuum, or pipes subject to
external water pressure, must be designed to withstand the differential
pressure by their own stiffness. Tests on PE pipe under hydrostatic
loading show deformation to proceed progressively until at about 10%
deflection when the rate of deformation increases and can lead to
total collapse ( Jenkins and Krolle, 1981). This 10% deformation has
been taken as the point of failure. This study also showed strong support
for the Gaube et al. form of correction factor cg as in equation (22.13c).
22.6 Long-term buckling pressures — constrained pipelines

Where pipes were buried in compacted soil or gravel, the collapse
resistance was increased. Performance of flexible pipe is determined
as much by the quality and density of compaction of the soil around
the pipe as by the properties of the pipe itself. If the buried flexible
pipe is subject to internal vacuum, soil will resist the collapse. The
mechanism of buckling will be an increase in any initial deformation
of the pipe. The amount of support provided by the soil could be
calculated approximately by mathematical methods and corresponding
formulae have been proposed. For polyethylene pipes at temperatures
below 278C (808F), the chance of a pipe failure due to internal
vacuum is virtually impossible in adequately compacted soil surround-
ings.
Where the pipeline is laid in a trench and backfilled, the deformed
shape of pipe may be significantly influenced by the nature of the back-
fill and bedding material and on the manner of placement. When the
pipe is buried and able to take advantage of support from surrounding
soil or grout media then the pipe should resist a differential pressure
greater than that given by equation (22.11).
Allman (1975) offered an expression for buckling pressure in
constrained soil which has often been used in the absence of other data:
p
pk ¼ 0:67 ðE0 pc Þ ð22:14Þ
where pk is the buckling pressure in constrained soil, pc is the un-
constrained buckling pressure from equation (22.11) including any
reduction for the deformed shape, and E0 is the tangent modulus of
the soil ¼ 12 er, e being Spangler’s (1941) modulus of passive soil
454
Flexible pipe
Table 22.3. Correction factor fc v.
(%) 1 2 3 4 5 6
fc 0.91 0.84 0.76 0.7 0.64 0.6
resistance and r the pipe radius. The modulus of soil reaction E0 is not a
constant but among other factors depends upon the stiffness of the pipe
used. Spangler recommended that E0 values should be based upon
experience and judgement.
Equation (22.14) is acknowledged to be a relatively crude tool for
calculating collapse pressure.
Compacted soil embedment factors in the pipe zone increase pipe
resistance to hydrostatic collapse. For properly compacted embedment
areas a supporting factor fos ¼ 3 may be applied. Taprogge (1981)
calculated critical pressure pc using the relationship:
pc ¼ po fc fos ð22:15aÞ
where po is the buckling pressure for a round pipe, fos is a factor of
safety ¼ 3 for properly compacted embedments, and fc is an ovality
correction factor. Table 22.3 lists values of fc as a function of and
shows how fc varies with deflection.
For polyethylene sewer pipes Taprogge indicates a factor of safety
ð fosÞ ¼ 1:5 so that the allowable differential pressure becomes:
pex p ¼ pc =1:5 ð22:15bÞ
For GRP pipes the vertical deflection is limited to 5%.
Greatorex (1981) has supplied data for GRP/RPM pipes and some of
his comments were as follows.
(a) In the USA GRP pipes are supplied with stiffness EI=D3 ¼ 1280 N/m2
and this is considered close to the optimum for most applications
with a 5% deflection being normal. In Europe GRP pipe is supplied
with a range of stiffness.
(b) Pipes of stiffnesses < 1000 N/m2 cannot be safely installed at any
depth where there is a risk of full vacuum occurring.
(c) Minimum stiffness with 2D cover should be the greater of
1100 N/m2 or that obtained from the equation:
p
pex p ¼ f32E0 rtEI=D3 g1=fos ð22:16Þ
where pex is external pressure; p is the internal pressure (ve if
vacuum); E0 rt is the tangent value of the resultant modulus of
455
Table 22.4. E0 tr v. modified Proctor density
Modified Proctor density 80 85 90 95
E0 tr (MN/m2 ) 3.0 5.0 7.0 10.0
the embendment including native soil ¼ 1.5E0 r; EI=D3 is the long-

term elastic stiffness of RPM pipe, taken to be 2/3 the initial stiff-
ness; and fos is the factor of safety. Greatorex (1981) used
fos ¼ 3 for RPM but others have used fos ¼ 2. Some values of
E0 tr are included in Table 22.4 as a function of modified Proctor
density.
3:0 E0 r 20:0 MN=m2 for normal embedments
Including the factor of safety ð fosÞ ¼ 3, initial pipe stiffness and E0 r,
equation (22.16) becomes:
p
pex p ¼ f3:56E0 rEI=D3 g ð22:16aÞ
(d) Minimum stiffness for 1:0 D depth of cover ¼ 1500 N/m2 .
(e) Minimum stiffness for installation with < 1 D depth of cover is
given by the lowest value from the equation using alternative
values of m:
pex p ¼ Et=ðrðm2 1Þf1 þ ½mL=ðrÞ2 gÞ þ EI=½ð1 2 Þr3
ðm2 2 þ ð2m2 1 Þ=f1 þ ½mL=ðrÞ2 gÞ
ð22:17Þ
where r ¼ Dm=2, L is the length between pipe sockets (assuming
relatively stiff sockets), Et is the tangent modulus of passive soil
resistance, and m is the number of waves ¼ 1, 2, 3, . . .
(f ) With depth of cover 1:6 D the pipe may buckle under vacuum
by heaving soil upwards if the pipe is held rigidly at its sides. Pipes
are more likely to deform upwards by sidefill compaction during
installation and the deformation may remain under shallow
cover, especially in the absence of vehicle loading.
For SDR 18 pipes the collapse pressure was 2 unsupported value;
for SDR 32 pipes the collapse pressure was 3 unsupported value;
and where grout fill was used, collapse pressure increased by five to
six times provided any gap in the grout was <1 or 2 D in length.
456
Flexible pipe
Kurt and Mark (1981) quoted the equation of Slocum for collapse
pressure of thermoplastic elliptical sections as:
pc ¼ pb 9 ð22:18Þ
where ¼ Dmin =Dm ¼ 1 .
At 5% eccentricity ¼ 0:05 and ¼ 0:95, giving p=po ¼ 0:63%,
and at 10% eccentricity ¼ 0:10 and ¼ 0:90, giving p=po ¼ 39%.
These results do not consider the stiffening effect of the reducing
minimum radius of curvature of the deforming section.
Investigations have shown that it is possible to place pipes in clay
surround and the necessary interaction between pipe and soil will be
obtained. However, the support from this clay will be less than if the
trench were refilled with sand. The pipe has to be designed for its
surroundings.
Leaving aside the matter of factor of safety, the equation for long-
term buckling pressure pex p is:
p
pex p ¼ f32E0 trEI=D3max g ð22:16Þ
This may be written in terms of the original round pipe diameter Dm
thus:
p
pex p ¼ ð1 "f Dm =sÞ1:5 f32E0 trEI=D3m g ð22:19Þ
For long-term buckling calculations either the long-term ‘creep’
modulus E50 may be used giving stiffness ¼ E50 s3 =ð12D3m Þ. Alterna-
tively the long-term elastic stiffness ¼ 0:7Eo s3 =ð12D3m Þ can be used.
The factor 0.7 is an adjustment for reduction in modulus over time
and may also include some compensation for out-of-roundness. The
long-term modulus is thus taken as 70% of the short-term modulus.
Equation (22.19) would indicate that if:
1 "fð50Þ Dm =s ¼ 0
then the pipe will buckle under a small differential pressure pex p, giving:
"fð50Þ ¼ s=Dm ¼ Dfð50Þ ð50Þ s=Dm
or
Dfð50Þ ð50Þ ¼ 1; giving ð50Þ ¼ 1=Dfð50Þ ð22:20Þ
In the long term Dfð50Þ may reach values of 10—15, corresponding with
values of ð50Þ in the range 6—10%. Practical experience has shown
that thin-walled pipes have collapsed after some years, even at low
backfill heights — that is, low values of pex p.
457
For thicker-walled pipes Dfð50Þ 3 to 4, showing that ð50Þ 0:25 or

25% for pipe to fail through instability. Prior to this happening however,
some thermosetting plastics such as GRP would fail due to cracking from
excessive bending strain. Soil movement would also be liable to occur.
22.7 Deformation of a circular section and its effect on wave

speed
Changes of pipe cross-sectional area due to alteration in shape are often
relatively small in comparison with the original undeformed area. In
considering the overall change of area with changing pressure, the
influence of hoop stress included in the majority of everyday pressure
transient analyses is assumed independent of the shape-change
component and these two effects are considered additive.
The act of increasing ovalisation or of partial re-rounding produces a
change in the cross-sectional area of the pipe regardless of any change in
the circumference. The term:
1=A dA=dp
which appears in the equation of conservation of mass then consists of
two components. First, there is the rate of change of area due to
changing hoop stress in the pipe wall but without shape change, as
included in elastic theory based waterhammer or pressure transient
computations. Second, there is a rate of change of area with changing
shape caused by changing internal pressure. Both of these components
will influence the rate of propagation of a pressure wave through the
composite liquid/pipe medium.
To illustrate the effect of changing shape, consider a deformed shape
having the properties:
perimeter ¼ ða þ bÞ ð22:21Þ
where a and b are the semi-major and semi-minor axes as shown in
Fig. 22.3.
Cross-sectional area ¼ ab ð22:22Þ
Neglecting the effect of hoop strain on the pipe circumference,
assuming perimeter length to be constant,
2r ¼ ða þ bÞ
if
a ¼ r þ h and b ¼ r þ v ð22:23Þ
458
Flexible pipe
Pex
d
s
P
Dm
Pex Pex
P P
d s
Pex
Fig. 22.3. Definition sketch for deformation analysis
where h is the horizontal radial deviation from the circular shape and
v is the corresponding vertical deviation, then if circumference is of
constant length:
2r ¼ ðr þ h þ r þ vÞ ¼ 2r þ ðh þ vÞ
giving,
h ¼ v ¼ ð22:24Þ
Substituting in the formula for cross-sectional area A then:
A ¼ ab ¼ ðr þ Þðr Þ ¼ ðr2 2 Þ ð22:25Þ
Irrespective of the value and sign of the cross-sectional area is
smaller than for the circular form with ¼ 0.
Differentiating with respect to :
dA=d ¼ 2
and
1=A dA=d ¼ 2=ðr2 Þ ¼ 2=r2 ð22:26Þ
From theory of elastic stability described by Timoshenko (1936), the
relationship between deflection and buckling pressure is given by:
=i ¼ pb =ð pb pex þ pÞ ð22:27Þ
459
where differential pressure normal to the pipe wall is pex p, pex being
the external pressure and p is the internal pressure; is deformation and
i is the initial deformation. pb is the differential pressure which
produces buckling in a circular section and is given by:
pb ¼ 2E=ð1 2 Þðs=Dm Þ3
or
hb ¼ 2E=½ gð1 2 Þðs=Dm Þ3 ð22:11Þ
where hb ¼ pb =ðgÞ. When pex p ¼ pb the pipe collapses completely.
Writing:
¼ i pb =ð pb pex þ pÞ
and differentiating with respect to p:
d=dp ¼ i pb =ð pb pex þ pÞ2
Assuming this is applicable to the approximately elliptical shape
under consideration, then:
1=A dA=dp ¼ 1=A dA=d d=dp ¼ 2=r2 ½i pb =ð pb pex þ pÞ2
or
1=A dA=dp ¼ 2fi pb =rg2 =f pb pex þ pg3 ð22:28Þ
In considering the speed of propagation of a pressure wave in an
elliptically deforming HDPE outfall pipeline, Larsen (1976) derived
the expression:
ðA Ao Þ=Ao ¼ 3fi pb =½rð pb pex þ pÞg2
Differentiating with respect to p gives:
1=A=dA=dp ¼ 6fi pb =rg2 =f pb pex þ pg3 ð22:29Þ
Equations (22.28) and (22.29) have different coefficients, reflecting the
alternative assumptions with respect to the shape of deformed section
but the form of expression is the same.
Whereas the expression for changing area, at constant shape,
due to hoop stress yields a constant gradient 1=A dA=dp ¼ Dc1 =ðsEÞ,
the gradient of area with pressure under changing shape is itself
pressure dependent. The pressure wave propagation rate is not a
system constant but varies with changing flow conditions in a similar
manner to inclusion of gas/vapour content variation with pressure.
460
Flexible pipe
For simplicity, ignoring the effect of gas/vapour, acoustic wave speed

a now becomes:
p
a ¼ 1= ð f1=K þ Dc1 =ðsEÞ þ c2 ðhb Þ2 =½ gðhb hex þ hÞ3 gÞ
ð22:30Þ
with pressure terms p, pb and pex having been replaced by the corre-
sponding pressure head values h, hb and hex , using h ¼ p=ðgÞ.
The changing shape factor effectively creates more storage within the
pipe than is given by simply using the area change due to hoop stress
variations. The effect is to produce a reduction in wave speed with
falling pressure and therefore a reduced pressure wave amplitude or
‘damping’ of the wave. Figure 22.4 shows the dependence of wave
speed upon ovalisation for HDPE pipes of SDR ratio 17.6 and 26.0
and for initial deformations i of 2% and 4%. Variation of the
coefficient c2 from 2 to 6 was also included. SDR 11 pipes are
sufficiently stiff to display little variation of wave speed with internal
pressure.
Larsen (1976) predicted the propagation of pressure transients
using the irregular grid of characteristics produced by varying wave
speed caused by ovalisation. While quite suitable for the relatively
straightforward case of a single pipeline without internal boundary
conditions or branches, this approach becomes more awkward for
general use. Just as the option exists of separating air or gas from the
water component when predicting wave propagation following air/gas
release under low pressures, so the same approach can be used to
separate the ovalisation factor from the area change produced by
hoop stress changes. This may be achieved by ‘collecting’ the ovalisa-
tion effect at computing sections in the same way as gas or vapour is
numerically collected. Alternatively the ovalisation effect can be
analysed as a separate pressure-dependent component distributed
along the pipeline.
Larsen (1976) was primarily interested in the use of HDPE pipes for
undersea outfalls and his concerns were largely with pipes surrounded
by water. In this context he presented curves showing how theoretical
buckling pressure varied when stiffeners in the form of anchoring
weights were placed at intervals along an HDPE pipeline. Inclusion of
pipe ovalisation would also be applicable to above-ground installations
or buried pipe where the response of the pipe was not significantly
influenced by its surroundings. This implies soft soil surroundings
such as marsh, swamp and loose or slightly compacted fill. Influence
461
210 SDR = 17.6

c2 = 6
200 Di = 0.02
190
SDR = 26
180 c2 = 2
Di = 0.02
SDR = 17.6
170
c2 = 6
Di = 0.04
160
150
140
Acoustic wavespeed a (m/s)
130
SDR = 26
120 c2 = 6
Di = 0.02
110
100
90 SDR = 26
c2 = 6
80 Di = 0.04
70
60 HDPE pipes
50
40
30
20
10
0
–10 0 10 20 30 40 50
Internal pressure head (mWG)
Fig. 22.4. Wave speed as a function of deformation
of any backfill is ignored in unconstrained conditions. Where a

compacted backfill is involved, the properties of fill have an important
influence on behaviour but the short-term effect of pipe ovalisation on
buckling pressure could also be included as outlined in the following
section.
462
Flexible pipe
22.8 Short-term elastic buckling under hydraulic transient

effects
22.8.1 Unconstrained conditions

If the pipe is installed in air, water or in very soft soils such as wet clay
which provide little support, critical pressure becomes:
pk ¼ 24:0 0:7Eð0Þ s3 =ð12D3m Þ=ð1 2 Þ
¼ 2ð0:7Eð0Þ Þðs=Dm Þ3 =ð1 2 Þ ð22:31Þ
22.8.2 Constrained conditions

If flexible pipes are exposed to a sudden negative pressure, the risk of
buckling can be calculated according to the following formula, provided
that the pipes are surrounded by granular soil which can give elastic
support to the pipe.
In studies of GRP pipe, Carlstrom (1981) introduced a reduction
factor for the stabilising effect of the soil modulus into the equation
for critical pressure, giving:
p
pk ¼ ð1 "f Dm =sÞ1:5 f32E0 t0:7Eo s3 =ð12D3m Þg ð22:32Þ
where ¼ 1 p=ð p þ pexð50Þ Þ, p being the transient vacuum pressure
within the pipe and pexð50Þ the long-term vertical pressure on the
pipe. Long-term pipe stiffness ¼ 0:7Eo s3 =ð12D3m Þ has been adopted as
the pipe may be more prone to buckle in the longer term when subject
to the same pressure transient. Eo is the initial flexural circumferential
modulus.
pexð50Þ ¼ Cgs B
where s is the weight of soil/m3 and Table 22.5 provides some typical
values of s . Cg may be obtained from Fig. 22.5.
Carlstrom (1981) considered friction between the soil prism and the
surrounding fill as one component and friction between the native
material forming the trench wall and the neighbouring fill as a second
Table 22.5. s v. soil type
Soil type Granular/ Stony/ Wet— Clay/heavy Water-

non-cohesive sandy loamy loam saturated clay
s =water 1.7 1.9 2.0 2.1 2.2
463
Cg
1 1.5 2 3 4
40
30
a
b a
c
d b
e c 2
20
d
e
15 1.5
10 1
9 0.9
8 0.8
7 0.7
H/B
6 0.6
H/B
5 0.5
4 0.4
3 0.3
2.5 0.25
2 0.2
1.5 0.15
0.10
0.1 0.15 0.2 0.3 0.4 0.5 0.6 0.8 1.0 1.5
Cg
Fig. 22.5. Cg as a function of depth of cover/trench width
component r , so that the vertical weight of overburden was modified

by a factor r . He also considered that over time the vertical pressure
would increase and so he proposed use of the factor ð1 þ Þ=
2ð1 þ r Þ=2 in compensation for long-term effects so that vertical
pressure was given by:
Pexð50Þ ¼ ð1 þ Þ=2ð1 þ r Þ=2ð ps þ pt pwu pw Þ
where ps is the pressure from overburden on the pipe, pt is pressure due
to traffic loading, pwu is uplift pressure due to groundwater, and pw is
internal water pressure.
464
Flexible pipe
Much of the foregoing is based upon the assumption of a generally

elliptical deformation with y=h ¼ 0:91.
A number of studies have been carried out of flexible pipe defor-
mation behaviour using finite-element analysis of the soil structure.
The possibility exists of coupling hydraulic transient analysis with the
numerical analysis of soil behaviour to provide a more detailed picture
of deformations.
22.9 Application of a flexible pipe model

Some examples have been included to illustrate the effects of pipe
flexibility upon the propagation of pressure transients. Different types
of simple pipeline configuration have been considered.
22.9.1 Long horizontal pipeline

The first case considers a long horizontal pumping main. Overall
length of the pipeline is 12.5 km and an overburden or external
loading equivalent to 2.5 mWG has been included.
Pipe bedding was included which offered some lateral support to a
deforming pipe. Initial ovality was set at 2% and gas release at low
pressure was included. The suction well level was set at þ0.0 mWG
and discharge elevation was þ1.0 mWG. Figure 22.6 shows predicted
Horizontal flexible pipeline. con = 1.0, del/D = 2%, h b = 12.3, gas included
70
Series 1
60 Series 2
Series 3
50
Elevation (mAD)
40
30
20
10
0
0 1000 2000 3000 4000 5000 6000 7000 8000 9000 10 000 11 000 12 000
–10
Chainage (m)
Fig. 22.6. Envelope curves for horizontal pipe and 2% deflection ratio
465
Horizontal flexible pipeline. con = 1.0, del/D = 2%, h b = 12.3

50
Series 1
Series 2
40
30
Head (mAD)
20
10
0
2.46
19.68
36.90
54.12
71.34
88.56
105.78
123.00
140.22
157.44
174.66
191.88
209.10
226.32
243.54
260.76
277.98
295.20
312.42
329.64
346.86
364.08
381.30
398.52
415.74
432.96
450.18
467.40
484.62
–10
Time (s)
Fig. 22.7. Head variation for horizontal pipe after pump failure
minimum head following a pumping failure. Maximum head is that of

steady pumping. In Fig. 22.7 changing head at the pumping station
and approximately halfway along the main are shown. A significant
amount of damping is evident in these results. If gas release is ignored
and the pipe has the same trench support as before, the minimum head
along the main after pump trip falls to a greater extent and buckling
occurs at chainage 11.5 km as depicted in Fig. 22.8. The model was
Horizontal flexible pipeline. con = 0.9, del/D = 2%, h b = 12.3, no gas

70
Series 1
60 Series 2
Series 3
50
Elevation (mAD)
40
30
20
10
0
0 1000 2000 3000 4000 5000 6000 7000 8000 9000 10 000 11 000 12 000
–10
Chainage (m)
Fig. 22.8. Envelope curves showing prediction of pipe buckling
466
Flexible pipe
set to terminate computations in the event of a pipe collapse. Where

the buckling limit is below the gas release level then the formation of
gas- or vapour-filled cavities will inhibit any further fall in head and
so internal pressure should not reach the level at which buckling occurs.
22.9.2 Descending outfall

The second example considers an idealised outfall similar to one of the
case studied by Larsen (1976). The pipeline descends uniformly over a
distance of 3 km. Pumping is used to overcome resistance with the
suction well level set at þ0.0 mAD. The pipeline is at 30.0 mAD at
its discharge point. An initial uniform eccentricity of 2% was adopted.
Gas release was represented and the pipeline was without lateral
support. After pump trip the minimum head shown in Fig. 22.9 was
predicted. The damping effect shown is essentially due to the pipe flex-
ibility as positive internal pressures preclude gas release. Appreciable
damping of the pressure wave is also evident in this figure.
If the pipeline is supported by stiffeners, flexibility is reduced. After
trip, the minimum head along the main falls to a greater extent than
before, as shown in Fig. 22.10. Greater amplitude of head variation
also occurs than without supports, again indicating a smaller amount
of damping in this stiffer pipeline.
Descending flexible pipeline. con = 6, del/D = 2%, h b = 12.3, gas c = 0.000 01

40
Series 1
30 Series 2
Series 3
20
Elevation (mAD)
10
0
0
100
200
300
400
500
600
700
800
900
1000
1100
1200
1300
1400
1500
1600
1700
1800
1900
2000
2100
2200
2300
2400
2500
2600
2700
2800
2900
3000
–10
–20
–30
Chainage (m)
Fig. 22.9. Envelope curves for descending outfall after pump failure and 2%
deflection ratio
Descending flexible pipeline. con = 1.0, del/D = 2%, h b = 12.3, gas c = 0.000 01
40
Series 1
30 Series 2
Series 3
20
Elevation (mAD)
10
0
0
100
200
300
400
500
600
700
800
900
1000
1100
1200
1300
1400
1500
1600
1700
1800
1900
2000
2100
2200
2300
2400
2500
2600
2700
2800
2900
3000
–10
–20
–30
Chainage (m)
Fig. 22.10. Envelope curves for descending outfall after pump failure and with
stiffeners
22.9.3 Uniformly rising main

A further example studied a uniformly rising pipeline of length 3 km.
This was also one of the configurations examined by Larsen. If the pipe-
line is supported along its length then following a pump trip, minimum
head is able to fall to a considerable extent as shown in Fig. 22.11. If the
Rising profile. Flexible pipeline. con = 1.0, del/D = 2%, h b = 12.3

35
Series 1
30 Series 2
Series 3
25
Elevation (mAD)
20
15
10
0
0
100
200
300
400
500
600
700
800
900
1000
1100
1200
1300
1400
1500
1600
1700
1800
1900
2000
2100
2200
2300
2400
2500
2600
2700
2800
2900
3000
–5
Chainage (m)
Fig. 22.11. Rising main, envelope curves with pipe stiffeners
468
Flexible pipe
Rising profile. Flexible pipeline. con = 6.0, del/D = 2%, h b = 12.3

35
Series 1
30 Series 2
Series 3
25
Elevation (mAD)
20
15
10
0
0
100
200
300
400
500
600
700
800
900
1000
1100
1200
1300
1400
1500
1600
1700
1800
1900
2000
2100
2200
2300
2400
2500
2600
2700
2800
2900
3000
–5
Chainage (m)
Fig. 22.12. Rising main, envelope curves for unsupported pipe
pipeline is unsupported, damping due to shape change is greater and

minimum head after trip does not fall to the same extent as before,
as illustrated in Fig. 22.12. If a rigid pipe is assumed, the minimum
head after pump failure is determined by vapour cavity formation, as
illustrated in Fig. 22.13, while if gas release and vapour cavity formation
Rising profile. Rigid pipeline. con = 0.0, del/D = 2%, h b = 12.3

35
Series 1
30 Series 2
Series 3
25
20
Elevation (mAD)
15
10
0
0
100
200
300
400
500
600
700
800
900
1000
1100
1200
1300
1400
1500
1600
1700
1800
1900
2000
2100
2200
2300
2400
2500
2600
2700
2800
2900
3000
–5
–10
Chainage (m)
Fig. 22.13. Rising main, envelope curves for rigid pipe and gas release
469
Rising profile. Rigid pipeline. No air content. con = 0.0

35
Series 1
30
Series 2
25 Series 3
20
15
Elevation (mAD)
10
0
0
100
200
300
400
500
600
700
800
900
1000
1100
1200
1300
1400
1500
1600
1700
1800
1900
2000
2100
2200
2300
2400
2500
2600
2700
2800
2900
3000
–5
–10
–15
–20
Chainage (m)
Fig. 22.14. Rising main, envelope curves for rigid pipe and no gas release
are both ignored, the minimum head falls to a much greater extent, as
depicted in Fig. 22.14.
The previous two examples considered produced predictions closely
in line with Larsen’s results when similar pipe properties were used. This
finding would indicate that it is quite feasible to utilise a fixed grid
computational scheme for such studies as well as a variable wave
speed scheme.
22.9.4 Pipeline of differing properties

Many pipeline networks are made up of a set of pipelines, each of which
has different properties. If flexibility of a pipe is considered then even if a
pipe of uniform characteristics is considered — that is, constant diameter,
material and wall thickness — a radical change in the pipe support
arrangement may be sufficient to produce a different response from
individual sections of a main. An outfall system may comprise a landward
section of pipeline which is laid in a trench with well-compacted backfill
and a seaward section which is essentially unsupported. The landward
stretch of main is unlikely to deform to the same extent as the seaward
element under changing pressure.
As an example, consider a pumped outfall having a horizontal land-
ward pipeline section of length 3.0 km and invert level þ1.0 mAD,
470
Flexible pipe
30
d/s PS
Seawall
25 Outfall
20
15
Head (mAD)
10
–5
–10
Time (s)
Fig. 22.15. Composite system, head variations after pump trip
leading to an embankment or seawall at level þ2.5 mAD. A seaward

section extends for a further 3.0 km, initially falling at a gradient
1 :200 and then more steeply at 1 :100 to a final discharge level of
14.0 mAD. Pumping station suction well level was at 3.0 mAD.
After pump trip, head falls steeply at the pumping station and for an 8 s
interval Fig. 22.15 shows rarefaction waves propagating along the
landward section with little damping evident, producing a minimum
pressure of around 8.0 mAD. Once over the seawall and into the
descending outfall section, noticeable damping occurs. Envelope curves
for the 6 km main show most clearly the effects of the different responses
of the two sections of main. The relatively stiff landward section allows a
substantial fall in head after trip. Within the seaward section, damping
inhibits the development of low pressures to an appreciable extent
(Fig. 22.16).
The cases considered were primarily used to demonstrate the
feasibility of carrying out computations for deforming pipelines under
changing transient pressure. Including what appears to be a significant
effect within day-to-day investigations, as a matter of routine is con-
sidered worthwhile. Obtaining the necessary information to allow
471
30
Invert lev.
Max. head
25
Min. head
20
15
10
Head (mAD)
–5
–10
–15
–20 Chainage
Fig. 22.16. Composite system, envelope curves following pumping failure
such computations to be undertaken remains an area of particular

interest.
22.10 Cyclical oscillations

Kirby (1981) concluded that brittle failure of PVC pipes was the most
important operational factor caused by pressure transients. These
failures appeared to be particularly associated with pumped sewerage
installations. Unlike many water supply systems, sewage pumping can
be more intermittent with relatively frequent starting and stopping of
pumps and substantial pressure waves produced by closure of non-
return valves. This leads to a rapid accumulation of large-amplitude
oscillations. In laboratory tests, Gotham and Hitch (1975) found that
if maximum amplitude of these oscillations was 50% of the rated
pipe pressure, any need for reduction in hoop stress safety factor was
avoided.
Vinson (1981) found that repetition of large-amplitude pressure
surges in PVC pipelines could lead to failure even where the peak
transient pressure remained below the pipe burst pressure. Depending
upon the applied stress, the pipes could sustain millions of loading
472
Flexible pipe
cycles before failure. A log relationship was established between hoop

stress and the hydrostatic life of PVC.
For 150 mm diameter PVC pipes with SDR of 17 and 26, the number
of cycles C to failure was:
C ¼ 3:7 1021 4:898 ð22:33Þ
where is peak hoop stress ¼ pD=ð2sÞ in psi.
The relationship above represents the lower 95% confidence limit of
the measured data.
Where the pipe wall has been scratched to a depth of 9% or more of
wall thickness, fatigue life is reduced and an effective wall thickness ¼ s
— scratch depth should be used when computing hoop stress. This
equation is a tool for use in large-amplitude conditions where peak
transient pressure > normal maximum working pressure.
Fedosoff and Szpak (1979) studied the effects of cyclic overpressures
in HDPE pipes. Their studies into ‘creep’ in this material showed that
when subject to cyclic overpressures of 2.5 times the normal design
stress, there was evidence of strain hardening in the material. They
concluded that if a ‘critical strain’ limit of 6—7% was not exceeded,
there was no permanent damage to the pipe.
473
23
Amplification of transient
pressures
The potential for actual transient pressures to exceed the expectations

predicted by a straightforward analysis has been demonstrated with
respect to the presence of modest pockets of air or gas within a pipeline
system. Where a network of pipelines occurs in which a range of pipe
sizes is present there also exists a potential for transient pressures to
increase in amplitude as pressure waves propagate through the
system. Some examples will be considered to illustrate circumstances in
which this can occur. Before considering actual systems within which
pressure transient amplification has been observed, three instances will
be examined which have the potential to produce transient pressure
wave increase.
23.1 Transmission of pressure waves through a branch

connection
These relatively abrupt rarefaction and compression waves developed
on the suction side of a booster pumping station, when pumps are
respectively started and stopped, propagate upstream along the trunk
suction main. Components of these pressure waves will travel into
any branch connections made from this main.
Considering Fig. 23.1 and assuming for simplicity that the trunk main
of area At supplies only the pumping station with water at a velocity Vo
and that zero flow enters the distribution main of area At . Further,
assume negligible pipeline resistance and a constant speed of pressure
waves. At each connection, the head at the connection point can be
calculated using the equation:
X X
H ¼ 1=g ðA JÞ= ðA=aÞ
474
Amplification of transient pressures
Distribution mains
At Ad
Suction trunk main Ad

Vo
J+ = Vo2/(1 + 1/2Ad/At)
+T
J– = Vo(1 – 1/2Ad/At)/(1 + 1/2Ad/At) PS
J– = 0
J+ = +Vo
Rising main
+X
J– = –Vo
+X
Fig. 23.1. Definition sketch for connection analysis
or
X X
H ¼ a=g ðA JÞ= ðAÞ ð6:4Þ
where wave speed a is considered constant. Suppose a pump trip occurs
at the PS which reduces the initial velocity Vo to zero. The resultant
invariant travelling upstream from the pumping station will have a
value Vo . The invariant from the upstream conduit has value þVo
and the value from the trunk main is zero. Head at the connection
point is then:
H ¼ a=gVo 2At =ð2At þ Ad Þ
where At is the cross-sectional area of the trunk main and Ad is the
cross-section of a distribution main.
Velocity in the trunk main upstream of the connection becomes:
Vt ¼ Vo Vo 2At =ð2At þ Ad Þ
and the resultant invariant propagating upstream of the connection
is:
Jt ¼ Vo ð1 12 Ad =At Þ=ð1 þ 12 Ad =At Þ ð23:1Þ
475
Velocity at the distribution main is:

Vd ¼ Vo 2At =ð2At þ Ad Þ
and the invariant travelling into the distribution main is:
Jd þ ¼ Vo 2=ð1 þ 12 Ad =At Þ ð23:2Þ
when the ratio Ad =At < 1 the invariant travelling into the distribution
main will have a larger value than the invariant arriving at the branch
connection.
23.2 Pressure wave transmission through a change of

cross-section
As pressure waves travel into the distribution network there is a
tendency for the pipe sizes to become smaller as the more remote
parts of the network are encountered. If a pipe is connected to a
second pipe of small diameter (Fig. 23.2), then head H at the con-
nection will be given by:
H ¼ 1=gðA1 J1 A2 J2 Þ=ðA1 =a1 þ A2 =a2 Þ
Assuming for simplicity that a1 ¼ a2 ¼ a and that initial flow is at
rest V ¼ 0 under head Ho ¼ 0. Let a compression wave travel in pipe
H1
+a
Ho = 0
A1
A2
V1 V=0
+X
J2+ = (J1+)2A1/(A1 + A2)
J2– = 0
J1+ = V1 + g/aH1
Fig. 23.2. Definition sketch for analysis of cross-section change
476
1 towards the connection with an invariant value ¼ J1 . Then:

H ¼ a=gðA1 J1 A2 J2 Þ=ðA1 þ A2 Þ
¼ a=gð J1 A2 =A1 J2 Þ=ð1 þ A2 =A1 Þ ¼ a=g J1 =ð1 þ A2 =A1 Þ
Velocity in the smaller pipeline 2 is given by:
V2 ¼ J2 þ g=aH ¼ þJ1 =ð1 þ A2 =A1 Þ
The invariant value propagating into pipe 2 is then:
J2 ¼ 2J1 =ð1 þ A2 =A1 Þ or J2 =J1 ¼ 2=ð1 þ A2 =A1 Þ ð23:3Þ
The invariant in pipe 2 has a greater value than that in pipe 1 while
A2 =A1 < 1:0.
Suppose a compression wave propagates in the positive direction
towards the connection of two pipes producing a velocity change of
1 m/s and amplitude change 100 m behind the wave front. Further,
assume velocity and head in front of the wave to be zero. Let acoustic
wave speed a in both pipes be 1000 m/s and for simplicity let accelera-
tion due to gravity g be 10 m/s2
J1 ¼ V1 þ g=aH1 ¼ 1:0 þ 10:0=1000:0 100 ¼ 2:0 m=s
and
J2 ¼ V2 g=aH2 ¼ 0:0 10:0=1000:0 0 ¼ 0:0 m=s
then
H ¼ 1000:0=10:0 ðA1 2 A2 0Þ=ðA1 þ A2 Þ
¼ 200=ð1 þ A2 =A1 Þ
If A1 ¼ A2 then H remains at 100.0 m and V remains at 1 m/s in the
downstream pipe after passage of the compression wave. If A2 < A1
then H > 100:0. Substituting the value for H into the expression for
J2 then:
J2 ¼ 0 ¼ V2 10:0=1000 200=ð1 þ A2 =A1 Þ
giving
V2 ¼ 2:0=ð1 þ A2 =A1 Þ
and again if A2 < A1 then V2 > 1:0 m/s. The pressure wave travelling
downstream in pipe No. 2 will have a quasi-invariant value:
J ¼ V2 þ g=aH ¼ 2:0=ð1 þ A2 =A1 Þ þ 0:01 200=ð1 þ A2 =A1 Þ
¼ 4:0=ð1 þ A2 =A1 Þ
477
When A2 < A1 then J > J1 and so the pressure wave travelling into a
network containing pipes of reducing diameter may be amplified. For
example, if A2 ¼ 12 A1 then H ¼ 133:0 m/s and V2 ¼ 1:33 m/s, giving
a transmitted wave invariant value ¼ 2.67 m/s as compared with the
incident wave value of 2.0 m/s.
23.3 Meeting of opposing pressure waves

Pipe networks serving distribution areas can also contain looped pipe
arrangements. Individual pressure waves travelling into the system
along branch connections with the trunk main may eventually meet
within the distribution system. Consider a pair of opposing compression
waves of equal invariant value travelling within a pipe as illustrated in
Fig. 23.3. When the waves meet, the solution at the meeting point can
H H
+a –a
+V V=0 –V
+X
J+ = V + g/aH J– = –V – g/aH
–a aV/g +a
H H
+V V=0 –V
Fig. 23.3. Meeting of opposing pressure waves
478
be obtained from:
V ¼ ð Jþ þ JÞ=2 and H ¼ ð Jþ JÞ=ð2g=aÞ
with velocity reduced to zero and head H ¼ a=gV. The head rise is
increased by the intersection of opposing compression waves. When a
pair of opposing rarefaction waves meet, the head drop at intersection
of the waves is likewise greater, as will be illustrated in section 23.5
concerning wellfield transients.
23.4 Pressure waves in a suction main

The first example of increased pressure wave amplitude comes from a
relatively simple treated water system in which a branch connection
is created halfway along an existing gravity main. A booster pumping
station is introduced on the new branch main a short distance down-
stream of the connection to the gravity main. Two duty pumps and a
standby unit were to be installed at the new pumping station. The
study was carried out to establish conditions along the cement-lined
ductile iron branch main which has a length of over 8 km and is ND
300. The gravity main is relatively short at 1700 m and is also of ductile
iron with diameter ND 600. The study was primarily concerned with
the new rising main and its protection, with information on the new
main and its profile, design flow rate and booster pump characteristics
readily available. When it came to the existing system, data were
more difficult to access and it would have been tempting to simplify
by assuming a range of constant heads on the suction side of the booster
station. This head range could have covered the normal operational
head range at the connection point.
23.4.1 Protection of the rising main

Studies showed that a pressure vessel was required to protect the rising
main against sub-atmospheric pressures following a station blackout
affecting all operational pumps. The resultant envelope curves for the
main were as shown in Fig. 23.4. The range in head just downstream
of the pumping station was around 80 m.
23.4.2 Conditions in the gravity main

Within the 17 900 m long upstream gravity main which was modelled in
detail, the corresponding envelope curves for the pump failure event
479
350
300
250
Elevation (mAD)
200
150
100
Invert
50 Max. head
Min. head
0
Chainage
Fig. 23.4. Envelope curves for rising main on pump start
were predicted to be as illustrated in Fig. 23.5 for the circumstance

when inlet valves at the downstream end of the main were shut. The
increased amplitude of oscillation over the lower parts of the main
can be seen with head range comparable to that in the rising main
and the appearance of sub-atmospheric pressures over parts of the
pipeline. Figure 23.6 shows time histories of head after pumping failure,
just upstream and downstream of the booster pumping station and at
the downstream closed valve at Rawyards over an interval of 1 min.
300
250
200
Elevation (mAD)
150
100
Invert
50
Max. head
Min. head
0
Chainage
Fig. 23.5. Envelope curves for gravity main on pump start
480
350
300
250
Elevation (mAD)
200
150
100
Rawyards
50 d/s pump
u/s pump
0
Time (s)
Fig. 23.6. Head variations in the gravity main after pump start
The amplification of surge amplitude is quite evident at this down-

stream closed valve.
This simple case highlights the need to include all parts of a system
which may be significantly influenced by transient effects. The capacity
of pressure vessel necessary to protect the rising main would not be
greatly altered by simplifying conditions upstream — that is, by imposing
a constant upstream head at the pump suction. Adverse conditions
within the existing gravity main would not have been identified by
such a measure. It is very desirable to include, as far a possible, existing
elements of the system and the range of possible flow conditions
therein, even if full details cannot be obtained due to passage of time.
23.5 Amplification within a network
23.5.1 Kirkleatham Lane Pumping Station

Treated water is supplied to both residential and industrial areas on
Teesside through an extensive network of trunk mains as shown in
Fig. 23.7. Industrial users include a steelworks which receives a bulk
supply through a DN 250 steel main branching from conduit No. 62
(C62) which is 600 mm in diameter. A motorised valve is used to
control flow into the steelworks tank. As part of a plan to provide
treated water to areas of the North York Moors, a pumping station
was developed at Kirkleatham Lane (Fig. 23.7) at the downstream
481
Longnewton TWL = 61 mOD
C53 840f C67 1016f
Uplands PS
Tank or reservoir
Pumping station
C62 600f conduit No. and diameter

C53 C67
5 km
C77 900f C67

C67
300f
Maltby Grange
C53
TWL = 63 mOD
C61 760f
Booster PS 250f
C66 381f C53 C61
Hemlington Park C42 450f

TWL = 46.65 mOD C46 760f
C48 C45
Longbank 450f 760f
TWL = 84.15 mOD C56 530f
Nunthorpe C47
380f C43
380f C51 450f
Booster PS
Steel 250f
C69 C62 600f
South Lakenby TWL = 50 mOD 914f
BS tank
Kirkleatham Lane PS
Redcar
distribution
zone
AC 600f
Yearby TWL = 85.675 mOD
Fig. 23.7. Distribution system leading to Kirkleatham Lane pumping station
limit of C62. From this pumping station a rising main extends to Yearby
Reservoir at an elevation of 85.675 mAOD. From Yearby a series of
pumping stations and pumping mains serves substantial areas of the
North York Moors. Between the steelworks branch and Kirkleatham
PS lies the Redcar residential area served by a set of branch connections
482
250f to High St West 150f

BS Redcar
Esplanade 150f
2 ¥ 150f
Coatham Rd High St 150f
228f
Corporation Rd
C62 600f from 150f
South Lakenby
Trent Road 200f
457f
Kirkleatham Lane PS
Troutbeck Rd West Dyke Rd
457f 228f
150f
Roseberry Road Race
600f AC to 150f course
Yearby Res.
150f 380f 150f
457f 150f
Coast Rd
Redcar Lane 150f
Low Farm Drive 380f
150f 150f
150f
200f
Laburnum Rd
Kirkleatham Lane West Dyke Road 150f
250f 228f
250f
0 1 2 km
Fig. 23.8. Redcar distribution area (simplified)
from C62. Figure 23.8 shows the networks of pipelines supplying Redcar
as modelled in this study. Branches from C62 comprise two DN 150
mains along Corporation Road, a DN 457 main laid along Trent
Road and a DN 250 pipeline into Kirkleatham Lane itself.
Following earlier studies, a pressure vessel was installed on the down-
stream side of Kirkleatham Pumping Station to protect the rising main
to Yearby from sub-atmospheric pressures. On pump start-up there is an
upsurge on the downstream side of the pumps and a drawdown on the
suction side. The presence of the pressure vessel on the delivery side of
the pumps allows a relatively smooth and gradual rise in pressure to
occur. In contrast, on the suction side the drawdown is very abrupt.
Similarly, when Kirkleatham pumps are switched off there is a
smooth and controlled decline in the head downstream of the pumps
483
110
100
Head (mAOD)
90
80
70
0.00 0.25 0.5

Time (min)
Fig. 23.9. Head downstream of Kirkleatham Lane pumping station after pump trip
regulated by the pressure vessel (Fig. 23.9), while on the suction side
pressure rises sharply (Fig. 23.10).
23.5.2 System modelling

With an extensive network of large-diameter trunk mains to consider,
it might be tempting to ignore or greatly simplify the network of
smaller-diameter distribution pipes. For example, residential demands
90
80
70
Head (mAOD)
60
50
40
0.00 0.25 0.5

Time (min)
Fig. 23.10. Head upstream of Kirkleatham Lane pumping station after pump trip
484
might be represented by a single drawoff. While it may be reasonable

to suppose that the transient response of the distribution area will
not have important consequences for the trunk main system as a
whole, the converse is not necessarily true. Surge behaviour within
the network of larger mains can have an important influence upon
behaviour within the smaller mains of the distribution system.
23.5.3 Recorded pipe bursts and pressure extremes

Within the distribution system of Fig. 23.8 a number of bursts were
recorded. Figure 23.11 shows head and flow just upstream of the
Metres
50 * * * * *
40
30
20
10
*Pressure surge at pump stop exceeds 50 metres

Time
0
30 Kirkleatham PS
outflow – MLD
3 bursts 1 burst
20
10
Time
0
Fig. 23.11. Head recordings at Kirkleatham indicating pipe bursts had occurred
485
90
80
70
60
Flow (l/s)
50
1
40
30
20
3 bursts burst
10
0
hr:mn 00:00
MT/94 26/10 27/10 28/10 29/10 30/10 31/10 01/11 02/11 03/11 04/11
Time
Fig. 23.12. Time history of velocity
pumping station. When head rises and flow falls to zero this indicates
that pumps have been tripped. Three bursts were recorded on one
occasion within the distribution zone and a further single burst after
a subsequent pump trip. Flow recordings with the distribution system
(Fig. 23.12) noted an abrupt flow increase which correlated with the
pipe bursts. Figure 23.11 shows the predicted steep head increase
upstream of the pumping station following pump trip, with a
maximum head of around 86 mAOD. Within the distribution system
along Redcar Lane, the corresponding predicted head variation is
shown in Fig. 23.13, with maximum head exceeding 150 mAOD.
Minimum head was also predicted to fall below atmospheric pressure,
which may interfere with sensitive equipment such as that used for
home dialysis which draws supplies direct from mains.
Predicted peak pressure within the distribution system, following
trip of Kirkleatham pumps, is shown in Fig. 23.14 in the form of
486
160
Head in Redcar Lane

140
120
Head (mAOD)
100
80
60
40
0.00 0.25 0.5
Time (min)
Fig. 23.13. Time history of head
contours of maximum piezometric level. These show a steady increase

in maximum pressure with distance into the network from the trunk
main. Maximum pressure within the system can exceed about 150%
of the peak pressure developed on the suction side of the pumping
station. These peak pressures correlated with the observed pipe bursts
which occurred in the network.
This example serves to illustrate that when a transient initiated
within a larger diameter of pipe is able to propagate into a system of
smaller pipes then the potential exists for the amplitude of pressure
fluctuations to increase substantially. In the case examined, one
remedy studied was to include a pressure vessel on the suction side of
the pumping station. This vessel attenuated the upsurge following
pump trip and allowed the effects of wave reflections within the
distribution system to reduce the severity of both maximum and
minimum pressures. The potential for pressure waves created at the
steelworks tank inlet valve was also considered, with the simplest
remedy here being control of valve stroke time in order to produce
sufficiently gradual rates of flow change and thus modest pressure
gradients.
487
12.0
11.0 13.5
13.0 13.9
10.1
10.6
10.7
14.0
14.5
10.8
10.5 15.0
12.5 10.6
15.5
12.9 10.8
14.0
10.9
13.0 15.7
12.0
11.0
11.0
12.0
11.1
11.2
11.5
11.3
11.4
Fig. 23.14. Contours of maximum pressure following pump trip
23.6 Wellfield transients
23.6.1 Collector pipeline system

The collector pipeline systems of many wellfields consist of pipes having
different diameters. The flow cross-section of each pipeline in the
network will be related to the number of wells contributing to that
pipe and their combined flow. It will therefore be the case that smaller
pipes will be used in the parts of the network more remote from a
receiving reservoir. Figure 23.15 shows the plan arrangement of an
extensive wellfield in the United Arab Emirates (UAE). Pipe diameter
488
W-111
Reservoir compound W-105 W-103
DN 600 DI
W-106
W-113
Motorised valves
Isolating valve ‘A’ Non-return valve
Line 2
Line 3
Line 4
Line 1
Line 5
Isolating valve ‘B’

Line 6
W-49
Line 1
DN 100 DI
W-1 W-124
W-125
W-123
Wellhouse
0 1 2 3 km
Fig. 23.15. UAE wellfield configuration
489
close to the reservoir was 600 mm while at the extremities of the system,
for instance near to well number 1, diameter was 100 mm.
23.6.2 Borehole and wellhouse configuration

A typical wellhouse arrangement may be as illustrated in Fig. 23.16.
In this case a check valve was located directly on top of the multi-
stage borehole pump, thus maintaining the riser full of water when
the pump is idle. In many instances underground water is being
mined with little recharge to replenish the store of water. Over time
a substantial variation in aquifer water level will occur, not only due
to the development of a cone of depression around each well but
also as a consequence of the long-term depletion of the resource
(Fig. 23.17). The delivery head where flow leaves the wellhouse is a
function of the receiving reservoir level and also depends upon flows
from other operating pumps contributing to the same system of
collector pipelines. The often substantial change in pump operating
head requires some means of regulating flow. A flow regulating valve
can be fitted in the wellhouse to ensure that short-term variations in
discharge are restricted to design values.
Longer-term lowering of water level in the aquifer may require pumps to
be uprated to accommodate the increased lift. Additional impellers may
be added to individual pumps, with the motor having been chosen to
suit the anticipated maximum number of stages. These measures to
control maximum discharge from a borehole are designed to avoid over-
pumping which could damage the well. Control of discharge also has
beneficial effects as far as pressure transient behaviour is concerned. It
should be noted that not all wellfields include wellhead flow regulation.
Where the well is able to accommodate the flow developed when aquifer
level is at its maximum without damaging the well then any transient
investigation should consider this peak flow condition.
An air valve may also be included at the wellhead and the possible
operation of this valve during a transient event should be considered.
23.6.3 Wellfield operating conditions

The wellfield illustrated in Fig. 23.15 is intended to operate as two
separate pumping systems. The isolating valves A and B are normally
closed, with all wells connected to collector main No. 1 known as
GROUP I and these wells feeding into collector mains No. 2, No. 3,
No. 4, No. 5 and No. 6 known as GROUP II. In emergency
490
Control valve
electrically operated
Compensator
Air valve
Flow meter
Flap valve Gate valve
Pressure
gauge
Ground level
Branch pipeline
Annular concrete plug

Stand pipe
Riser pipe from pump
Gravel packing
Casing
Non-return valve
Multi-stage pump
Motor
Aquifer
Aquifer
Sump pipe
Clay (impervious)
Bail bottom
Fig. 23.16. Typical well arrangement
circumstances when one of the DN 600 mains downstream of valve A is

out of service, this valve can be opened to allow flows from both groups
of pumps to reach the reservoirs via a single DN 600 line. When a
collector main is isolated upstream of point A then valve B is opened
491
Borehole losses
Control valve loss
Collector
main loss
Minimum
Pumping head
pumping head
after some time
Reservoir
Wellhouse
Initial water level
Cone of depression
Static water surface

after some time
Cone of depression
Aquifer
Bore hole
Fig. 23.17. Head conditions at a well
to permit flows from one group of pumps to pass into the alternative
system of mains in order to reach the reservoirs. Normally, around
50% of the pumps would be in use at any time, yielding about 20 Mld.
23.6.4 Pumpset inertia

The moment of inertia of the borehole pumpsets can often be quite
small. An equation for moment of inertia (MI) for a multi-stage
pump is of the form:
MI ¼ Motor inertia þ stage inertia number of stages ð23:4Þ
For the wellfield shown, pumpset inertia ranged from 0.0255 kg/m2 to
0.0314 kg/m2 . With such low inertia it may be expected that a very
rapid fall in head downstream of the wellhead will occur following a
492
pump trip and that the wellhead and other air valves may come into
operation.
23.6.5 Sequenced pump operation

Start-up of pumps is under operator control and to avoid excessive
demand upon the electrical supply network, sequential start of pumps
is desirable with a time delay between successive pump operation.
Considerable choice exists in the sequence used to start pumps. Opera-
tional circumstances may dictate that a certain set of pumps be used but
that may still leave the starting sequence in the operator’s hands. The
mode of pump starting can have an important effect upon the
maximum transient system pressure. For the system being considered,
some general guidelines can be established by considering a small
number of pump start sequences from the large number of possibilities.
Alternatives are illustrated using the pumps of GROUP I. Figure 23.18
shows the predicted piezometric head at well No. 1, located at the
upstream end of line No. 1 for a starting arrangement whereby pumps
of GROUP I were operated in sequence with a 1 s time delay between
operating successive pumps. Pumps were started at well No. 1 with the
final pumps to be started being those nearest to the reservoirs.
Maximum head predicted was below 380 mAD.
If the reverse sequence is used — that is, first starting those pumps
nearest to the reservoirs and finally the pumps around well No. 1 —
380
370
360
Elevation (mAOD)
350
340
330
320
310
0.00 0.25 0.5 0.75 1.00

Time (min)
Fig. 23.18. Head at well No. 1 for sequenced start commencing at well No. 1
493
420
400
Elevation (mAOD)
380
360
340
320
0.00 0.50 1.00 1.50

Time (min)
Fig. 23.19. Head at well No. 1 for sequenced start ending with well No. 1
somewhat higher peak pressures were predicted. Figure 23.19 shows the
corresponding head variation at well No. 1. Maximum transient pres-
sure exceeded 420 mAD. Other starting arrangements also produced
higher starting pressures than those of Fig. 23.18 and it was concluded
that the optimum arrangement was to first start those pumps towards
the upstream end of the system and progress downstream towards the
reservoirs.
23.6.6 Pumping failure

Pumping failure has to consider the effect of an area power failure
affecting the entire wellfield. Instead of a pressure transient being
developed at a pumping station and spreading along its rising main,
in the case of the wellfield with a collector pipeline system, failure of
pumps distributed throughout the network means that piezometric
level starts to fall almost simultaneously at each wellhouse. With the
low moment of inertia of borehole pumps used in installations of this
kind, the rate of fall in piezometric level is rapid at any single wellhead.
The presence of an air valve having a good inflow capacity will usually
ensure that sub-atmospheric pressures local to the well remain modest.
As the downsurge spreads into the collector network there is oppor-
tunity for interaction between each of the rarefaction waves shown in
Fig. 23.20, with the potential for amplification of the head drop
produced at individual wellhouses.
494
(i) Ho = steady pumping head
(i)
Vo Collector main
Vo
Wellhouse Vo 2Vo
Riser Branch
Pump (a)
Ho
(ii)
(ii) Ho
(ii)
(ii)
Ho
(iii)
(iii)
Vo
2Vo
Ho
(ii)
(b)
Ho
(ii)
(iv)
(iv)
(iii)
(iii)
2Vo
Vacuum pressure
(c)
Fig. 23.20. Head variations due to area blackout
495
360
340
320
mAOD
300
280
260
0 1 2 3 4 5 6
km
Fig. 23.21. Area blackout envelope curves along line No. 1 (initial stages)
Figure 23.20a shows the initial rarefaction waves (i) travelling from
each wellhead along the branch pipeline towards its connection with
the collector main.
In Fig. 23.20b the initial rarefaction waves have reached the
connection point with the collector. A partial reflection in the form
of a compression wave (iii) travels from the connection back towards
each wellhouse while rarefaction waves (ii) travel both upstream and
downstream along the collector from the connection points. These
rarefaction waves produce a fall in head within the collector main.
In Fig. 23.20c opposing rarefaction waves have met in the collector
main between the branch connections, leading to a doubling in ampli-
tude of head drop (iv). There is a strong possibility that severe vacuum
pressures will occur in the collector between adjacent wells.
Simulation of an area power failure affecting all GROUP I pumps
shows that such vacuum pressures develop along line No. 1 as shown
in Fig. 23.21. The profile of line No. 1 shows relatively low elevations
over the middle parts of the main with higher levels both towards
well No. 1, chainage 0 km, and also at the reservoir end of the pipeline,
chainage 6 km.
23.6.7 Air valve operation

Low pressure head is experienced throughout the network with air
valves operating at most wellheads. As time passes, flow towards the
496
440
420
400
380
360
mAOD
340
320
300
280
260
0 1 2 3 4 5 6
km
Fig. 23.22. Area blackout envelope curves for line No. 1 (later stage)
reservoirs ceases and then reverses while positive flows continue from
the higher parts of the system around well No. 1. Air pockets at
these more elevated wellhead air valves will continue to grow while
air is expelled from the lower parts of the system. Air removal through
these lower air valves is facilitated by continued positive flows from the
higher parts of the system around well No. 1 and also by reversed flow
from the reservoirs.
As lower air valves close, compression waves are developed and a
general rise in head is developed over these lower central parts of the
network as shown in Fig. 23.21.
With reversed flow from the reservoirs increasing and air valves
closing, the compression wave propagating upstream travels into
progressively smaller pipes, being amplified until it finally enters the
DN 100 pipes around well No. 1. Closure of final air valves coupled
with the strength of the compression wave produces a maximum
transient head of over 440 mAD at well No. 1 (Fig. 23.22).
23.7 Potential for amplification

These examples serve to illustrate the particular problems associated
with complicated pipe networks where the pressure transients are
travelling into pipes of decreasing diameter and where looped arrange-
ments exist. The effect of pressure waves or the same type, either
compression or rarefaction, travelling in opposite directions carries
the risk of important amplification when these waves meet. Risk of
497
amplification is greater where the pressure transient is relatively

steep. Then there is only limited time for the effects of pressure wave
reflection to alleviate this effect. Use of pressure vessels can be very
beneficial in limiting rates of pressure change at the source of the
hydraulic transient, thus allowing the benefits of wave reflection to be
realised.
498
24
Flow instabilities
In the large majority of water and sewerage systems, whenever a change

of flow state occurs, the action of pipeline resistance is sufficient to
cause the transient effects to be gradually dissipated thus allowing a
new steady flow state to be realised. On occasion, however, certain
features of a network may act to sustain an unsteady motion so that
a self-perpetuating oscillation is developed which does not decay over
time but may in some instances even increase in severity. Such a
condition of increasing amplitude of oscillation or resonance can be
very dangerous and must be avoided.
The phenomenon of resonance has been known for a considerable
time with experiments on the topic reported by Camichel et al. as
early as 1919 and with a presentation of the theory of resonance by
Jaeger in 1939. Not all instances of unstable behaviour result in
resonance but nevertheless can have undesirable consequences.
24.1 Types of oscillation

The theoretical period of elastic oscillation within a pipeline of length L is
4L=a and this is the fundamental period for a simple pipeline of uniform
characteristics. Harmonics can also develop having periods of oscillation
which are fractions of the fundamental period. In a pipeline consisting of
a series of pipes of differing properties such as diameter and wave speed,
partial reflections at junctions of different pipe types can cause the
oscillatory motion to have a different period called the apparent period.
For a dangerous resonant condition to develop there has to be an
energy input during each cycle of oscillation. As a simple illustration,
consider a child on a swing. At the optimum stage in the child’s
movement the parent gives the swing an additional push (energy
499
input) as a consequence of which the child is able to swing to a higher

elevation with each successive cycle. This is an example of a forced
oscillation as an external agent is present. Ultimately, safety and
other considerations will curtail the height (amplitude) which the
swing reaches. Unfortunately, there are no such automatic constraints
on hydraulic systems.
A close investigation of boundary conditions led to the discovery of
self-exiting or self-induced oscillations in hydraulic systems. For
example, valve vibrations have been known to produce resonances in
pressure tunnels which have caused linings to become cracked. Pulsa-
tions have been observed in pipelines where the periodic movement
of turbine runner blades past fixed blades (guide vanes) has produced
pressure fluctuations. The period of these pulsations and resulting
resonance has often been eliminated by altering the number of
runner blades thus changing the relationship between the pipeline
period and the period of the pulsations.
Cyclical motion of a valve which allows a variable outflow to occur is
accompanied by pressure fluctuations which may become amplified if
the connected pipeline system responds to the same period of oscillation.
Resonance can only occur if there is an exciter present. This can be in the
form of a forcing function which excites the system at some period.
A second way in which resonance can occur is if say a valve responds to
oscillations occurring within the system, by self-excitation. In such cir-
cumstances the system controls the oscillation, and the operation of a
boundary condition such as a self-acting valve is influenced by the system.
Amplitude of a resonance may ultimately be limited if minimum head
during an oscillation falls to near the discharge head downstream of a
valve, so that energy conditions are not significantly influenced by
the valve response. When the head difference across a valve becomes
small it does not matter if the valve is fully opened or only partly
open as the flow rate through the valve will be small irrespective of
how much the valve is open. Resistance effects may act to some
extent to produce a peak amplitude <2Ho where Ho is the static
head upstream of the valve, as shown in Fig. 24.1.
While the resonant phenomenon has been well documented and
understood for many years, examples still continue to occur. Gordon
(2006b) recently described an instance of resonant behaviour at a
28 MW impulse turbine hydropower plant in the Andes. Rather than
installing the more usual cable connection from the turbine governor,
to control movement of the spear valve, it was decided to use a chain
of small ball bearings forming a flexible rod. It was hoped to avoid
500
Flow instabilities
Maximum amplitude = 2Ho
Constant head
Static head = Ho
Self-acting valve
Fig. 24.1. Development of resonance
any stretching which might occur in a cable connection by using the ball
bearings. Unfortunately, a small amount of slack in the chain of ball
bearings was sufficient to confuse the governor about the precise
valve opening. This caused ‘hunting’ of the valve at a frequency
equalling the period of the penstock and produced a self-exiting
oscillation. Reverting to a cable connection eliminated the problem.
The impedance method of analysis can be used to predict the most
critical periods of oscillation in a pipeline network. It is not intended
to discuss this method but the interested reader can find a description
of the method in books, for example by Streeter and Wylie (1967) or by
Jaeger (1977).
To illustrate circumstances in which it is possible for flow within a
pipeline network to become unstable and fail to settle at a steady
operating condition, three examples have been included in this chapter.
The first illustration concerns a pumping installation and the second an
entirely gravity-driven system.
24.2 Pumping system — Glasgow East Main and Daer

network link
Figure 24.2 shows the plan arrangement of trunk mains serving south
and east parts of the city of Glasgow and its environs. Two systems
501
East Main
Ruchazie TWL 102.9
Cranhill TWL 111.7
762f
Cranhill PS
457f
Castlemilk Low Castlemilk PS
TWL 128.6 610f
Burnside PS
Castlemilk High 457f
TWL 171.5 Link Main DN 700
Distribution main
to Cathkin DN 200
Link Main DN 700
East Rogerton 546f Dechmont

TWL 213.4
TWL 130.4
546f
244f
546f 546f
244f 244f
Auchentibber Udston
TWL 208.8 TWL 181.4
Water tower
Ground tank
Pumping station 546f 546f
Pressure reducing valve
Isolating or fixed throttle valve
0 1 2 3 4 5
Approximate scale – kilometres
from Daer reservoir
Fig. 24.2. East Main and Daer transmission main
are involved, one being the downstream part of the East Main supplied
originally with water from the Loch Katrine system and the other a
gravity-driven network supplied from Daer reservoir to the south.
These usually independent systems operate at significantly different
piezometric levels, with the Daer gravity system around 100 m higher
than the East Main network. To improve security of supply a link
main has been constructed between the two systems. This link runs
502
Flow instabilities
from Burnside on the East Main to East Rogerton SR on the Daer

system. Any interruption of supply within the East Main system
would be compensated by inflows under gravity from the Daer system
with a pressure-reducing valve in the link main to control the flow
entering the East Main network. Should flows in the Daer network
be disrupted, water would be pumped from the lower East Main into
the higher Daer pipeline system.
24.2.1 Burnside booster pumping station

The link main booster pumping station was located at Burnside as
shown in Fig. 24.2. This new pumping station was fitted with pressure
vessels both upstream and downstream of the pumps. The suction side
vessel had a gross volume of 27.5 m3 and contained an air charge of
9.0 m3 at a head of 99.0 mAOD. On the downstream side of the
pumps the pressure vessel had a total volume of 16.0 m3 and the air
charge was again 9.0 m3 under the much higher head of 213.4 mAOD.
24.2.2 Hydraulic transient computations

Prior to introducing the new booster pumping station at Burnside,
computations were undertaken to establish the effects upon existing
installations. Along the section of East Main system shown are a
number of branches connected to local booster pumps which lift treated
water into storage in tanks and water towers. These pumps serve storage
chambers at Cranhill, Castlemilk Low and Castlemilk High and each of
these duty pumps could remain in operation while the link main booster
pumps were running.
24.2.3 Castlemilk Low pumping station

Particularly significant was the booster pump serving Castlemilk Low
tank. The performance curve for this pump is shown in Fig. 24.3
together with the normal duty point prior to introduction of the new
booster pumping station at Burnside. The pump performance curve
contains a zone of unstable operation. Unstable curves are those
which contain a relatively flat region of Q—H curve or even where H
increases with Q over part of the performance curve. Considering the
circumstance in which the new link main pump is in operation, the
flow rate in the East Main leading to the link main connection will
be increased, causing head loss in the East Main to rise. The effect
503
100
90
80
70
Generated head (m)
60 Duty point –
link main pumps off
50
Hmax
40 Castlemilk Low pumps

characteristic curve
WPL SDC 7/8 UNIGLIDE
30
20
10
Cut impeller
0
0 100 200 300
Flow rate (l/s)
Fig. 24.3. Castlemilk Low pump performance curve
will be to reduce the suction head at existing booster pumps along the
East Main and in particular at Castlemilk pumping station. The steady
operating head for these pumps will increase. Added to this increased
pumping duty at Castlemilk Low station are transient pressure varia-
tions produced when the new booster pump is operated.
24.2.4 Transient pressures

There is a transient pressure drop as the new pump at Burnside is
started, as shown in Fig. 24.4, causing the temporary operating point
at Castlemilk Low pump to move upwards into the unstable zone of
operation. When the necessary transient pumping head exceeds Hmax
(Fig. 24.2), flow falls abruptly towards zero but as it does so, suction
pressure starts to rise because of the deceleration produced. This rise
in pressure will tend to move the operating point back into the stable
zone and allow pumping to recommence, but as soon as this happens
and velocity starts to increase, suction head falls once more and the
operating point moves back into the unstable region of pump operation.
Added to this are transient effects reflected from other upstream
parts of the network. The resultant effect is that behaviour remains
504
Flow instabilities
without link main pump
with link main pump
Pumping head
without link
Transient drawdown main pump
Link Main PS
Additional head
+ with link main
pump running
East Main
Castlemilk PS
Fig. 24.4. Hydraulic conditions along final part of East Main
unstable with an irregular cyclical oscillation remaining as long as

pumps are in operation. Figure 24.5 shows the predicted behaviour
following start of the Burnside link main booster pump. Trip of the
booster pump at Burnside allows the operating point of Castlemilk
Low pump to move back to its original location with stable flow
restored.
75
Start of link main pumps
70
Head u/s of Castlemilk PS
65
Head (mAOD)
60
55
50
0.0 0.5 1.0

Time (min)
Fig. 24.5. Head conditions at Castlemilk after link main pump operated
505
110
100 Burnside pumps tripped
90
Head (mAOD)
Head upstream of Castlemilk PS
80
Burnside pumps operating
70
60
0 1 2 3 4
Time (min)
Fig. 24.6. Head conditions at Burnside after link main pump is started
24.2.5 Spread of unstable oscillations and consequences

Unstable behaviour is not confined to the vicinity of the pump affected
but in common with all hydraulic transient effects will spread through-
out the pressurised pipeline system. In Fig. 24.6 is shown the predicted
variation of head upstream of the Burnside link main booster station
with the booster pump initially in operation. The unstable oscillation
present is eliminated when the link main pump at Burnside is tripped.
While the unstable motion predicted did not produce extremes of
pressure outwith allowable limits, undesirable consequences can occur
as a result of such behaviour. Pulsations of flow can cause movement
of check valves with consequent accelerated wear of hinges. Flow
and head fluctuations at a pump will produce a corresponding oscilla-
tion in power to the pump. These effects on pumps and valves are
not confined to the unstable pump itself but affect all pumping stations
of the system.
24.2.6 Possible remedies

It is important when enhancing or modifying an existing scheme to
ensure that the original elements of the system will continue to
behave in an acceptable manner. If this is not predicted to be the
case, necessary measures should be taken to restore stable operation.
In this case, the Castlemilk Low pump was originally fitted with a cut
impeller and so it would be a relatively straightforward matter to install
506
Flow instabilities
a larger-diameter impeller within the existing pump casing. This larger

impeller will allow an increase in duty head while remaining in the
stable zone of operation (Fig. 24.2). The increase in pumping head
will also have implications with regard to power of the motor required
to meet the revised duty when the link main pump is operating.
24.3 Gravity flow system

The second example concerns a pipeline conveying treated water under
gravity from an upstream impounding reservoir with an overflow
spillway level at 359.1 mAOD. A treatment works sited 3 km below
the dam has an outlet piezometric level of 283.68 mAOD. An 838 mm
internal diameter mild steel (MS) pipeline of length 26.686 km conveys
water to a downstream storage reservoir having a water level of
183.87 mAOD. Almost midway along the pipeline, 13.175 km from the
treatment works is a break pressure chamber (BPC). The undulating
profile of the main is shown in Fig. 24.7.
300
M 283.68 mAOD
957.6 litres/s
260
BPC
240 1078.6 litres/s 957.6 litres/s
220
200
M
180
160
140
120 838.2 mm f steel main
100
80 838.2 mm f steel main
60
40
0 2 4 6 8 10 12 14 16 18 20 22 24 26 28
Chainage (km)
Fig. 24.7. Hydraulic conditions along gravity main
507
Float chamber
Overspill weir
M
Linkage
Throttled connection
between float
chamber and BPC
838 mm f Float
MS main in
2 No. 457 mm f
float operated valves
838 mm f
MS main out
Fig. 24.8. Break pressure chamber configuration
24.3.1 The break pressure chamber

Upstream of the break pressure chamber (BPC) the 838 mm main bifur-
cates into twin 457 mm diameter lines, each of which terminates at a Glen-
field and Kennedy H1 float actuated valve. These valves regulate inflow to
the BPC. Each valve is connected to a float, with each float contained
within a small chamber inside the BPC (Fig. 24.8). DN 150 and DN 200
openings link the float chambers with the main tank of the BPC which
has plan dimensions of 3.05 m 3.66 m over the normal water level range.
24.3.2 Head losses

Measurements of flow and head loss in the section of pipeline upstream
of the BPC gave values of ks from 0.44 mm to 0.46 mm over the flow
range 82.7 Mld to 93.2 Mld. For the section of pipeline downstream
of the BPC ks was evaluated as 0.6 mm at a flow rate 82.25 Mld.
24.3.3 Flow regulation

Inflow to the downstream storage reservoir is normally controlled by
float actuated valves but with a butterfly valve upstream for emergency
508
Flow instabilities
Actual phase 2
port shape
Initial phase 2
port shape
Phase 1 and initial phase 2

valve travel = 85.7
Actual phase 2
valve travel = 103.2
Phase 1 port shape
Phase Total port area – mm2

1 37 316 (100%) 11 581 (67%) 1652 (33%) open
2 initial 62 581 (100%) 20 387 (67%)
2 actual 157 419 (100%)
Fig. 24.9. Valve porting arrangements
closure duty. In the event that these downstream valves are closed,
flow into the pipeline below the BPC will fall, causing level in the
chamber to rise. The floats in the BPC will then move upwards,
progressively closing the twin H1 valves and throttling flow entering
the BPC. Each H1 valve contains a set of four openings or ports
through which flow passes. Shape of the ports can be altered to provide
a range of flow v. head loss characteristics to suit particular require-
ments while providing the necessary maximum design discharge
when fully opened. If discharge required should change over time
the port shape can be altered to suit. Phase I of the project had port
shapes as illustrated in Fig. 24.9, with a total flow area 373.16 cm2
when fully opened and with a valve stroke of 85.7 mm. Originally it
was intended that the port shapes would be altered to permit an
increased maximum flow in Phase II through a total ported area of
625.81 cm2 over the same stroke length. However, to maximise
discharge a considerably increased flow area totalling 1574.19 cm2
was actually created, with stroke length being increased to
103.2 mm. These port forms for Phase II are also shown in Fig. 24.9.
Travel of floats within the BPC for the final Phase II arrangement
was 762 mm — that is, from a minimum elevation of 231.058 mAOD
509
7
BPC inlet valve loss coefficient (kV)
as a function of valve stroke
Phase I porting
5
Initial Phase II porting

log10 (kV)
1
Actual Phase II porting
0
0 10 20 30 40 50 60 70 80 90 100 110
Valve stroke (mm)
Fig. 24.10. Head loss characteristics for valve
to a maximum of 231.82 mAOD. Crest level of the overflow weir

within the BPC was at 231.744 mAOD so that when water level
exceeds crest elevation, spillage over the weir commences and water
runs to waste.
The ported area is dramatically increased for the actual Phase II port
shape. A corresponding drop in head loss coefficient is achieved, as
shown in Fig. 24.10. Loss coefficient for the actual Phase II port con-
figuration remains relatively low up until the final 10% or so of closure
so that the flow deceleration will be relatively modest over the first
90% of the closure process. Only over the final 10% or so does flow
deceleration become important. Flow thus continues to enter the
BPC at a relatively high rate over most of the valve stroke.
510
Flow instabilities
24.3.4 H1 valve movement

The relatively rapid flow deceleration during the final stages of closure
produces a substantial inertial head rise above the H1 valves and this
head increase propagates into the upstream pipeline as a compression
wave which is eventually reflected back to the BPC from the treatment
works after about 30 s (2L=a) as a rarefaction wave, at which time a
relief of pressure occurs above the BPC.
During the interval while the pressure wave has been travelling in the
upstream pipeline, water has been spilling from the BPC over the weir.
This produces a gradual reduction in water level in the chamber
allowing partial reopening of the float valves. Due to the poor charac-
teristic of the float valves in the actual Phase II port configuration,
even a modest reopening of the valves can produce a substantial flow
increase into the BPC, leading to a falling head in the pipeline upstream
of the valves. This fall in head is reinforced by the relief of pressure
brought about by the rarefaction pressure wave reflected back from
the treatment works. The combined effect is to create a more severe
fall in head upstream of the valves than would be produced by either
influence acting alone. This low head is then propagated into the
upstream pipeline as a rarefaction wave which travels to the treatment
works before being reflected back to the BPC as a compression wave,
arriving after a further time interval of around 30 s. During this interval
the water level in the BPC has continued to fall towards the overflow
crest. The returning wave causes a head rise upstream of the valves
and a renewed flow into the chamber. Water level in the BPC increases,
the float valves close once more and water starts to spill over the weir
crest. The high rate of flow deceleration at the valves produces a further
compression wave which travels into the upstream pipeline.
With the passage of time the combined influence of the elastic
oscillation in the upstream pipeline and the fluctuation of discharge
through the float valves produces a tendency for the amplitude of
oscillation to persist and even increase, as shown in the prediction of
Fig. 24.11. Period of the oscillation in head upstream of the BPC is
about 60 s which corresponds with the fundamental period of the
upstream pipeline 4L=a. Variation of water level in the BPC was also
predicted to have a period of about 1 min. This behaviour is an example
of a ‘self-excited’ oscillation. Figure 24.12 shows the predicted extremes of
head in the gravity main system over a 10 min interval following closure of
the inlet valves at the downstream reservoir. The amplitude of oscillation
is much greater in the section of main upstream of the BPC than in the
lower stretch of pipeline as a consequence of system excitation.
511
380
360
340
Head u/s of inlet valves at BPC

Head (mAOD)
320
300
280
260
240
0 1 2 3 4 5 6 7 8 9 10
Time (min)
Fig. 24.11. Head variations upstream of BPC
24.3.5 Remedial measures

Remedial measures were considered and one simple measure was to
ensure that the float valves were able to close fully. This was achieved
by increasing the height of the overspill weir within the BPC.
Eliminating spillage and the accompanying variations of water level in
the chamber allowed the valves to remain shut after initial closure.
Another measure considered was to alter phasing of float valve
response by introducing extra throttling between the float chambers
and the main tank of the BPC.
350 Maximum and minimum head

along gravity main
300 BPC
u/s reservoir
250
Head (mAOD)
200
150
d/s reservoir
100
50
Pipeline profile
Helvetica Flush left
0 2 4 6 8 10 12 14 16 18 20 22 24 26
Chainage (km)
Fig. 24.12. Envelope curves along gravity main
512
Flow instabilities
24.4 Small hydro station

The final example concerns behaviour observed at a small hydropower
station in New South Wales and which is operated by Snowy Hydro
Limited. Unstable behaviour has been observed since commissioning
of a submerged discharge valve (SDV) which is sited in parallel with
a hydro turbine. The hydraulic system consists of an intake tower
leading to a tunnel of nominal diameter 3750 mm and length 356 m.
At the downstream end of the tunnel is a trifurcation. Two of the
branches downstream of this trifurcation supply free discharge cone
valves of diameter DN 1800 and DN 1200. These valves were shut
during the events being considered. The final branch downstream of
the trifurcation is a penstock having nominal diameter 2000 mm and
length around 120 m. The penstock terminates at the small hydro-
station. The station houses the small hydro unit rated at 1.105 MW
and an SDV of the type shown in Fig. 9.13. Inlet to the SDV is
ND 800 with an ND 800 bend before a taper reducing diameter to
ND 600. Plan arrangement of the tunnel and penstock system is
shown in Fig. 24.13. When fully open the valve is capable of discharging
3.5 m3 /s. Flows from the SDV and from the small hydro unit enter
individual sumps with flow leaving each sump via an overflow weir
and entering a common outflow channel. The small hydrostation and
outflow channel can be seen in Fig. 24.14.
24.4.1 Observed behaviour

Observations of hydraulic transient behaviour within the penstock and
small hydrostation were reported by O’Connor (2007). Pressure
Small hydro turbine

1.105 MW
Intake tower
Penstock, 2000f Length 120 m
Submerged discharge
valve 800 ¥ 600
Trifurcation
Tunnel 3750f Length 356 m Free-discharge valves,

DN 1200 and DN 1800
Fig. 24.13. Small hydro pipeline system
513
Fig. 24.14. Turbine/valve house and outflow channel
measurements were made within the ND 2000 penstock both with and
without the small hydro turbine in operation. When flow through the
SDV was 2.7 m3 /s or higher, the following effects were observed:
(a) a pulsating sound within the small hydrostation
(b) the noise of cavitation from within the SDV in the area of the
DN 800 bend (Fig. 24.15)
(c) increased vibration in the small hydrostation
(d) more pronounced turbulence in the SDV sump evidenced by
surface wave action (Fig. 24.16)
(e) large pressure fluctuations recorded on pressure gauges at a hydrant
close to the penstock at the small hydrostation.
Recordings of pressure fluctuations in the penstock showed a
maximum head range of about 9 mWG. While a range of frequencies
of oscillation was recorded, it was noted that some oscillations had a
period of around 1.6 s which is close to the fundamental period 4L=a
of the conduit system as a whole. Other much higher frequencies
were also present.
514
Flow instabilities
Fig. 24.15. Submerged discharge valve bend
24.4.2 Comments on observed behaviour

In considering unsteadiness in pressure conduit systems Jaeger (1977)
concluded that no pipeline network is entirely safe against develop-
ment of harmonic vibrations and that there is no reason to conclude
Fig. 24.16. Turbulence in SDV sump
515
that any system is also safe from possible resonance developing.

Efforts of designers have been directed towards suppressing any possible
source of incipient vibration from valves, gates, guide vanes, etc.
Generally, the longer a conduit system the greater will be the amount
of damping present and the greater will be the amount of ‘exciting’
energy required to initiate development of an oscillation. In his
studies of resonance, Jaeger (1977) noted that oscillations were more
prone to occur in shorter systems with incidents reported for conduit
lengths of 698, 720, 1430 and 4214 m. The present system has a
relatively short length and also tunnel diameter was sized for operation
of the free discharge valves at greater flows than the SDV. Damping
in the system is therefore quite low. Configuration of branch serving
the SDV is also noteworthy. A step change in diameter from
2000 mm down to 800 mm is unusually large. A higher-frequency
pressure wave oscillation in the SDV branch will largely be reflected
from the large section change while longer period oscillations, say
with the overall system period of around 1.6 s, will not be affected in
this way.
Attention was focused on the SDV sump as a source of pressure
fluctuations. The radial jet issuing from the SDV will produce intense
shear and turbulence as it mixes with the surrounding water in the
sump. A state of continuous instability is created with a wide range of
frequencies of oscillation being present in the sump. Tritton (1980)
notes that large eddies have formed from what would seem to be
entirely insignificant perturbations or local instabilities in the mean
velocity profile. The life of these large eddies can be comparatively
short before a ‘normal’ flow is resumed Al Naib (1992).
24.4.3 Modelling behaviour

It was hypothesised that pressure fluctuations caused by vortex action
within the SDV sump caused by the radial jet motion might be
producing velocity and pressure oscillations at the base of the SDV.
These fluctuations would then be propagated at acoustic velocity,
through the valve and into the system as a whole. To examine the
possible effects of such transients a computer model of the system
was formed with the exit from the SDV subject to a range of frequencies
of oscillation. The frequencies used generally ranged from around 1/
40 s to 1.6 s. The higher frequencies represented ‘typical’ fluctuations
within the radial jet and the longer frequencies were from larger-scale
oscillations.
516
Flow instabilities
90
80
70
60
50
mAD
40
30
20
Trifurcation
SDV tee
10
SDV bend
0
Time (s)
Fig. 24.17. Head behaviour in SDV system
Using this approach it was possible to produce variations of head

of around 9 m in the penstock (Fig. 24.17). This amplitude was
maintained throughout much of the penstock and it is thought likely
that it is the longer-period oscillations that are responsible for the
pulsating sound observed. The longer-period oscillations were not
considered capable of creating cavitation in the SDV. The higher-
frequency components were largely reflected from the change of
cross-section (800 mm to 2000 mm). A large amplitude of oscillation
was predicted to occur in the vicinity of the SDV bend. Based on
averaged flow conditions across the pipe cross-section, cavitation was
not thought likely to occur. A more detailed consideration of flow
variation across the bend section was thought necessary. Ideally a
CFD analysis of the bend coupled with information from the transient
model of the system as a whole would be used to yield data on the 3-D
variation of flow and pressure at the bend. Due to limited time, a much
simpler approach was adopted to give some indication of the bend
effect. Rankine’s free vortex approximation was used to produce a
velocity variation across the bend from the outside radius to the
inside of the bend. Figure 24.18 shows a typical result with much
greater velocities and correspondingly lower pressures predicted close
to the inside wall of the bend. It was considered that a transient
cavitation might be occurring towards the inside wall of the bend
produced by a combination of low pressure during oscillations and
reduced pressure during curvilinear flow.
517
Gas charge connection
Diaphragm
Pulsation damper (accumulator)
Connection to SDV branch
DN 800
Flow
Low pressure towards

inside of bend
Velocity distrbution across bend

DN 600
Fig. 24.18. ‘Damper’ installation
24.4.4 Remedial measures

A number of remedial possibilities present themselves, ranging from
air injection to introducing increased compressibility to sump modifica-
tions to alter the nature of pressure fluctuations. Alteration of the
sump would require considerable time and expense, while air injection
would need to be more or less continuous and involve use of a
compressor.
The simplest remedy is to install a ‘pulsation damper’. This is in
essence a small pressure vessel fitted with a diaphragm and containing
a gas charge. Given that head room may be a problem, a relatively
flat pattern would seem appropriate, such as is illustrated in
Fig. 13.2c. The installation of this damper is shown in Fig. 24.18.
Sizing of the vessel requires computer simulation and a number of
studies were conducted to determine a lower limit of capacity. Figure
24.19 shows results of a simulation. Head variations in the penstock
(SDV tee) and at the SDV bend are shown to be quite modest for
the chosen gas charge of 5 litres. Overall capacity of the damper was
10 litres.
518
Flow instabilities
60.0
59.5
59.0
58.5
58.0
57.5
57.0
56.5
56.0
55.5
SDV tee
55.0 SDV bend
54.5
Time (s)
Fig. 24.19. Head variation in SDV system with damper
24.4.5 Final comments

This installation represents a fairly unusual SDV installation given the
large change of cross-section between the penstock and the SDV
branch. Coupled with the very low-capacity damping system as a
whole, this factor would seem to have allowed turbulent pressure
fluctuations to produce unsatisfactory behaviour within the system,
which would not have been manifest in other circumstances.
The potential for a small gas pocket to amplify transient effects has
been demonstrated in section 18.5 of Chapter 18. In conditions
pertaining at this SDV installation, a small pressure vessel has been
included to alleviate a local problem within a much larger system. It
is desirable to conduct analyses to ensure that no significant amplifica-
tion problems can arise within the system generally as a consequence of
including the small gas volume. In this connection studies of SDV
closure were carried out to verify that there was no risk of amplification
in this instance.
519
References
Allman, W. B. (1975) Operating Sections Proceedings. American Gas Associa-

tion Conference, Catalogue No. x-50875, May 5—8.
Bishop, R. R. and Lang, D. C. (1984) Design and Performance of Buried Fiber-
glass Pipes — A New Perspective. In Pipeline Materials and Design (Schrock,
B. J. (ed.)), pp. 1—12. New York: ASCE.
Camichel, C., Eydoux, D. and Gariel, M. (1919) Etude Théorique et Expérimen-
tale des Coups de Belier. Paris: Dunod.
Carlstrom, B. I. (1981) Structural Design of Underground GRP Pipe. Inter-
national Conference on Underground Plastic Pipe, 30 March to 1 April
(Schrock, B. J. (ed.)), pp. 56—78. New York: ASCE.
Colebrooke, C. F. (1939) Turbulent flow in pipes with particular reference to
the transition region between the smooth and rough turbulent pipe laws.
Journal of the Institution of Civil Engineers, 11, 133.
Collier, S. L. and Hoerner, C. C. (1983) A Facility and Approach to Performance
Test of Check Valves. New York: ASME paper No. 82-PET-8, Vol. 105, pp.
62—67.
Dawson, P. (c. 1980) The Design of Descending Pumped Pipelines with Particular
Reference to Effluent Disposal Sea Outfalls. Leeds: Hydraulic Analysis Limited.
Donsky, B. (1961) Complete pump characteristics and the effects of specific
speed on hydraulic transients. Transactions of the American Society of
Mechanical Engineers, Journal of Basic Engineering, December, 685—699.
Ellis, J. (1980) Study of Pipe—liquid Interaction Following Pump Trip and
Check Valve Closure in a Pumping Station. Proceedings of the 3rd Inter-
national Conference on Pressure Surges, Canterbury, 25—27 March. Paper
E3, pp. 203—220. Cranfield: BHRA.
Ellis, J. (1994) Hydraulic Instability in Water Supply Networks. Proceedings of
the 2nd International Conference on Water Pipeline Systems, Edinburgh, Scot-
land, 24—26 May, pp. 335—357. Cranfield: BHR.
Ellis, J. and Khairulla, L. M. (1974) Oscillations in a surge tank with upper and
lower expansion galleries. Water Power, November, 359—364.
520
References
Ellis, J. and Mualla, W. (1986) Selection of Check Valves. Proceedings of the 5th
International Conference on Pressure Surges, BHRA Fluid Engineering
Centre, Hannover, Germany, 22—24 September, pp. 213—222.
Ellis, J. and Tint, K. M. (1976) Simulation of pressure transients at valves and bifur-
cations. International Water Power and Dam Construction, 8:8, 24—27 August.
Esleek, S. H. and Rosser. (1959) Check Valve Water Hammer Characteristics.
Proceedings of an American Water Works Association Society meeting,
November.
Fedosoff, F. A. and Szpak, E. (1979) Cyclic Overpressure Effects on High Density
PE Pipe. Proceedings of Western Canada Water and Sewage Conference,
Regina, Saskatchewan. Engineering Digest, September, 35—38.
Fok, A. T. K. (1980) Design Charts for Surge Tanks on Pump Discharge Lines.
Proceedings of the 3rd International Conference on Pressure Surges, Eng-
land, March, BHRA, Paper J3, pp. 445—472.
Gaube, E., Müller, W. and Falcke, E. (1974) The Statics of Rigid Polyethylene
Drain Pipes. Proceedings of Kunststoffe, 64, April, 193—196.
Glass, W. L. (1980) Cavitation and Corrosion in a Pumping Main. Proceedings of
the 3rd International Conference on Pressure Surges, Canterbury, 25—28
March, Paper H3, pp. 401—414. Cranfield: BHRA.
Gordon, J. L. (2004) The untested computer program. Hydro Review World-
wide, 12:13, July, 32 and 34.
Gordon, J. L. (2006a) Nintendo engineers. Hydro Review Worldwide, 14:4,
September, 46 and 51.
Gordon, J. L. (2006b) The hunting impulse turbine. Hydro Review Worldwide,
14:1, March, 46 and 54.
Gotham, K. V. and Hitch, M. K. (1975) Design considerations for fatigue in a
PVC pressure pipeline. Pipes and Pipelines International, 20:2, February, 10.
Graze, H. R. (1968) A Rational Thermodynamic Equation for Air Chamber Design.
Proceedings of the 3rd Australian Conference on Hydraulics and Fluid
Mechanics, Institution of Engineers Australia, 25—29 November, p. 57.
Graze, H. R. and Forrest, J. A. (1974) New Design Charts for Air Chambers.
Proceedings of the 5th Australian Conference on Hydraulics and Fluid
Mechanics, Christchurch, New Zealand, December 9—13, Vol. II, p. 34.
Graze, H. R., Schubert, J. and Forrest, J. A. (1976) Analysis of Field Measure-
ments of Air Chamber Installations. Proceedings of the 2nd International
Conference on Pressure Surges, London, Paper K2. Cranfield: BHRA.
Greatorex, C. B. (1981) The Relationship between the Stiffness of a GRP Pipe and
its Performance when Installed. Proceedings of an International Conference
on Underground Plastic Pipe, New Orleans, 30 March—1 April (Schrock,
B. J. (ed.)), pp. 117—129. New York: ASCE.
Heinsbroek, A. G. T. J. and Tijsseling, A. S. (1994) The Influence of Support
Rigidity on Waterhammer Pressures and Pipe Stresses. Proceedings of the 2nd
International Conference on Water Pipeline Systems, Edinburgh, UK,
24—26 May, pp. 17—30. Cranfield: BHR Group.
521
Jaeger, C. (1939) Theory of Resonance in Pressure Conduits. Transactions of

American Society of Mechanical Engineers, February.
Jenkins, C. F. and Krolle, A. E. (1981) External Hydrostatic Loading of Poly-
ethylene Pipes. Proceedings of an International Conference on Underground
Plastic Pipe, 30 March—1 April (Schrock, B. J. (ed.)), pp. 527—541. New
York: ASCE.
Joukovsky, N. (1904) Basic waterhammer equations. Memoirs of the Imperial
Academy of Science of St Petersburg, 1898, 9:5. Translated by Simin, O.,
Proceedings American Waterworks Association, p. 341.
Keller, A. and Zielke, W. (1976) Variation of Free Gas Content in Water During
Pressure Fluctuations. Proceedings of the 2nd International Conference on
Pressure Surges, London, Paper D1, pp. D1-1—D1-12. Cranfield: BHRA.
Kirby, P. C. (1981) PVC Pipe Performance in Water Mains and Sewers. Proceed-
ings of an International Conference on Underground Plastic Pipe, 30
March—1 April (Schrock, B. J. (ed.)), pp. 161—174. New York: ASCE.
Koetzier, H., Kruisbruink, A. C. H. and Lavooij, C. S. W. (1986) Dynamic
Behaviour of Large Non-return Valves. Proceedings of the 5th International
Conference on Pressure Surges, Hannover, Germany, BHRA Fluid Engi-
neering Centre, 22—24 September, pp. 237—241.
Kot, C. A., Hsieh, B. J., Younghadl, C. K. and Valentin, R. A. (1980) Transient
Cavitation Phenomena in Fluid—structure Interactions. Proceedings of the 3rd
International Conference on Pressure Surges, Canterbury, UK, March,
BHRA, Paper E1, pp. 165—184.
Kurt, C. E. and Mark, K.-Y. (1981) Collapse of Noncircular Supported Thermo-
plastic Pipelines. Proceedings of an International Conference on Under-
ground Plastic Pipe, 30 March—1 April (Schrock, B. J. (ed.)), pp. 360—
372. New York: ASCE.
Larsen, I. (1976) Development and Propagation of Underpressure Fronts Influ-
enced by Buckling of the Pipe Wall. Proceedings of the 2nd International
Conference on Pressure Surges, London, 22—24 September, Paper J3, pp.
J3-43—J3-58. Cranfield: BHRA.
Linser, M. (2004) Innovative surge tank throttle — standing the test of time.
Hydro Review Worldwide, 12:4, September, 36—37.
Livingston, A. C. (1954) Reflux and Allied Self-closing Valves. Report No.
SP496. Cranfield: BHRA.
Livingston, A. C. (1969) Waterhammer control equipment applications. The
Glenfield Gazette, A Journal of Hydraulic Engineering, 229, February, 24—28.
Lupton, H. R. (1976) Opening Address. Proceedings of the 2nd International
Conference on Pressure Surges, London, England.
Manuel, A. R. (1970) Waterhammer in Pressure Conduits. Delft Hydraulic
Laboratory, Publication No. 54.
Martin, C. S. (1976) Entrapped Air in Pipelines. Proceedings of the 2nd
International Conference on Pressure Surges, London, Paper F2, pp. F2-
15—F2-28.
522
References
Martin, C. S., Padmanabhan, M. and Wiggert, D. C. (1976) Pressure Wave

Propagation in Two-phase Bubbly Air—water Mixture. Proceedings of the
2nd International Conference on Pressure Surges, London, Paper C1,
pp. C1-1—C1-16. Cranfield: BHRA.
McCrone, I. G. (undated) Surge in Pipelines. Kilmarnock: Glenfield & Kennedy
Limited.
Miller, E. (1969) The submerged discharge valve. The Glenfield Gazette, A
Journal of Hydraulic Engineering, February, 229, 17—23 and 229.
Misiunas, D., Lambert, M., Simpson, A. and Olssen, G. (2007) Assessing water
mains condition using hydraulic transients. Proceedings of the Institution of
Civil Engineers, Water Management, 160:WM2, June, 89—94.
Molin, J. (1981) Flexible Pipes Buried in Clay. Proceedings of an International
Conference on Underground Plastic Pipe (Schrock, B. J. (ed.)), 30 March—1
April, pp. 322—337. New York: ASCE.
Moser, A. P. (1981) Strain as a Design Basis for PVC Pipes? Proceedings of an
International Conference on Underground Plastic Pipe (Schrock, B. J.
(ed.)), 30 March—1 April, pp. 89—103. New York: ASCE.
Pake, M. C. (1983) Non-return Valve Testing: Description of a Rig and Preliminary
Results for a Double-disc Valve. Proceedings of the VI International Round
Table on Hydraulic Transients with Water Column Separation, IAHR, Glou-
cester, UK, 19—20 September. Marchwood: CEGB Engineering Laboratories.
Parmley, L. J. (1965) The Behaviour of Check Valves During Closure. Laboratory
report No. H64/51. Kilmarnock: Glenfield & Kennedy Ltd.
Perko, H. D. (1986) Check Valve Dynamics in Pressure Transient Analysis.
Proceedings of the 5th International Conference on Pressure Surges,
Hannover, Germany, BHRA Fluid Engineering Centre, 22—24 September,
pp. 229—235.
Petroff, L. J. (1984) Performance of low stiffness plastic pipe in stiff soils. Pipe-
line Materials and Design (Schrock, B. J. (ed.)), pp. 24—35. San Francisco,
CA: ASCE.
Pool, E. B., Porwit, A. J. and Carlton, J. L. (1962) Prediction of Surge Pressures
from Check Valves for Nuclear Loops. Paper No. 62-WA-219. New York:
ASME.
Popescu, M. (1985) Highly Efficient Protection Systems Against Waterhammer
for Pumping Stations. Proceedings of an International Conference on the
Hydraulics of Pumping Stations, Manchester, UK, 17—19 September,
pp. 77—88.
Provoost, G. A. (1980) The Dynamic Behaviour of Non-return Valves. Proceed-
ings of the 3rd International Conference on Pressure Surges, Canterbury,
UK, BHRA Fluid Engineering, 25—27 March, Paper No. J1, pp. 415—427.
Provoost, G. A. (1983) A Critical Analysis to Determine Dynamic Characteristic
of Non-return Valves. Proceedings of the 4th International Conference on
Pressure Surges, Bath, UK, BHRA Fluid Engineering, 21—23 September,
Paper F4, pp. 275—286.
523
Spangler, M. G. (1941) The Structural Design of Flexible Pipe Culverts. Iowa

State College Bulletin No. 30, December.
Taprogge, R. H. (1981) Large Diameter Polyethylene Profile-wall Pipes in Sewer
Applications. Proceedings of an International Conference on Underground
Plastic Pipe (Schrock, B. J. (ed.)), 30 March—1 April, pp. 175—190. New
York: ASCE.
Tucker, D. M. and Young, G. A. J. (1960) Estimation of the Size of Air Vessels.
Proceedings of the 7th Conference on Hydromechanics, College of Aero-
nautics, Cranfield, UK, 3—5 October, Paper SP670.
Tullis, J. P., Streeter, V. L. and Wylie, E. B. (1976) Waterhammer Analysis with
Air Release. Proceedings of the 2nd International Conference on Pressure
Surges, London, UK, 22—24 September, Paper C3, pp. C3-35—C3-47.
Cranfield: BHRA.
Vardy, A. E. (1976) On the Use of the Method of Characteristics for the Solution of
Unsteady Flows in Networks. Proceedings of the 2nd International Confer-
ence on Pressure Surges, London, UK, 22—24 September, Paper H2, pp.
H2-15—H2-30. Cranfield: BHRA.
Vinson, H. W. (1981) Response of PVC Pipe to Large, Repetitive Pressure Surges.
Proceedings of an International Conference on Underground Plastic Pipe
(Schrock, B. J. (ed.)), 30 March—1 April, pp. 485—494. New York: ASCE.
Whiteman, K. J. and Pearsal, I. S. (1959) Reflux Valves and Surge Tests at
Kingston Pumping Station. BHRA and NEL Joint report No. 1. Cranfield:
BHRA.
Wilkinson, D. H. (1980) Dynamic Response of Pipework Systems to Water
Hammer. Proceedings of the 3rd International Conference on Pressure
Surges, Canterbury, UK, BHRA, March, Paper E2, pp. 185—202.
Worster, R. C. (1959) The Closing of Reflux Valves. Report No. SP626. Cran-
field: BHRA.
524
Further reading
Abbott, M. B. (1966) An Introduction to the Method of Characteristics. London:

Thames & Hudson.
Ackers, P. (1963) Resistance of Fluids Flowing in Channels and Pipes, Hydraulics
Research Paper No. 1. Department of Scientific and Industrial Research,
Hydraulics Research Station. London: HMSO.
Al Naib, S. K. (1992) Jet Mechanics and Hydraulic Structures — Theory, Analysis
and Design. Romford: Research Books.
American Society of Civil Engineers (1975) Pipeline Design for Water and
Wastewater. Report of the Task Committee on Engineering Practice in
the Design of Pipelines. Philadelphia, PA: ASCE.
Andrews, D. H. and Kokes, R. J. (1963) Fundamental Chemistry. London: Wiley.
Bartlett, R. E. (1978) Pumping Stations for Water and Sewage. London: Applied
Science Publishers.
Biwater (1988) Water Industries Manual. Dorking: Biwater Limited.
Burstall, T. (1997) Bulk Water Pipelines. London: Thomas Telford.
Escarameia, M. (ed.) (2005) Air Problems in Pipelines, a Design Manual. Wall-
ingford: HR Wallingford.
Fox, J. A. (1977) Hydraulic Analysis of Unsteady Flow in Pipe Networks. London
and Basingstoke: MacMillan.
Hoechst Plastics (1971) Hostalen GM 5010 Pipes. Frankfurt, Germany:
Farbwerke Hoechst AG.
IAHER (2000) Hydraulic Transients with Water Column Separation, 1971—1991
Synthesis Report. Milan: International Association for Hydraulic Engineering
and Research.
Jaeger, C. (1977) Fluid Transients in Hydro-Electric Engineering Practice. Glasgow
and London: Blackie.
Knapp, R. T., Daily, J. W. and Hammitt, F. G. (1970) Cavitation. New York:
McGraw-Hill.
Linton, P. (1972) Note on Pressure Surge Calculations by the Graphical Method —
Pump Stoppage after Power Failure. TN 447, 3rd edn. Cranfield: BHRA Fluid
Engineering.
525
Miller, D. S. (1978) Internal Flow Systems. Volume 5 in the BHRA Fluid

Engineering Series. Cranfield: BHRA Fluid Engineering.
Miller, E. (c. 1970) Surge Protection of Pipelines. Internal report. Kilmarnock:
Glenfield & Kennedy Ltd.
Mualla, W. (1983) The Dynamic Behaviour of Check Valves. PhD thesis, Univer-
sity of Strathclyde, Glasgow.
O’Connor, J. (2007) Jindabyne Dam SDV Transient. Snowy Hydro Limited,
Cooma, NSW, Australia.
Smith, A. A. (1969) A Problem Oriented Library for Steady One-dimensional
Open Channel Flow. PhD thesis, University of Strathclyde, Glasgow.
Spalding, D. B. and Cole, E. H. (1963) Engineering Thermodynamics. London:
Edward Arnold.
Streeter, V. L. and Wylie, E. B. (1967) Hydraulic Transients. New York:
McGraw-Hill.
Thorley, A. R. D. and Enever, K. J. (1979) Control and Suppression of Pressure
Surges in Pipelines and Tunnels. Report No. 84. London: CIRIA.
Timoshenko, S. (1936) Theory of Elastic Stability. New York: McGraw Hill.
Tritton, D. J. (1980) Physical Fluid Dynamics. Wokingham: Van Nostrand
Reinhold.
526
Index
Note: Page number in italics refer to figures
parameter covers, 320321

conduit shape, 32 damage to components, 353
constraint factors, 31 definition sketches, 303, 305
fluid properties, 2728 DN 450 mains, 314316, 315317
gas influence on, 3238, 37 DN 700 mains, 314316, 315317
interpretation, 2740 envelope curves, 292, 310, 311
pipelines, 31 flow reversal, 302307, 303
sewage effect, 3840 flywheel case study, 169
simple expression, 2931 heads, 293, 303, 304305, 308310, 309,
tunnels, 31 311, 312
Abbott, -, 43 pump restart, 314, 315
abnormal behaviour pump start-up, 331
feeder tanks, 264267, 264, 266 small orifice, 316, 317
Abu Hamour PS 44 pumps, 180 hydraulic gradient, 317318, 318
accumulators, 334 in-line check valves, 406407
acoustic wave speed, 30, 37, 37 initial air volume, 306307, 307
actuated valves, 7177, 7274, 119122 inspection, 320
aerated jets, 373, 374 installation, 225229, 226229
air, 325358 large orifice, 310, 311, 314, 315
in pipelines, 314318, 315316, 317 liquid conveyance, 319320
pump blockage, 351354 locations, 297298, 297
pump start-up, 325339 materials, 321
removal, 314, 315316, 315, 316 operation, 299, 301
sewage rising mains, 345351 outflow devices, 332333, 343345
see also gas outlets, 292293, 292293
air compressors, 215216 pressure waves, 301, 302
air-filled risers, 325334 pump restart, 314318, 315316, 317
air flow pumping stations, 169, 310314,
double orifice valves, 299, 300 311313
air valves, 295324 reversed velocity, 306307, 307
after failure, 310, 311 second orifice type, 304
air removal, 315316, 316 sewage, 295, 296, 310314, 311313,
air volume, 306307, 307, 310, 311312 319
case study, 310314 Sharjah Pumping Station, 169
chambers, 320321 simulation uncertainties, 318319
closure, 308310, 309 small orifice, 316, 317
considerations, 318321 submersible pumping stations, 327, 329
527
air valves (continued ) back-flow connection, 282283, 283

surge alleviation, 295, 298302, 300, 301 backwash pumps, 244
venting, 317318, 318, 336 Bainbridge, -, 193, 194, 197, 198
wellfield operation, 496497 ball check valves, 424425, 425, 440
air venting Balmore
air valves, 317318, 318, 336 Loch Lomond scheme, 175
butterfly valves, 336, 338339 bearing friction moment
flow conditions, 330 valves, 411
sparg line, 339 bifurcation conditions, 8284, 82
air volume bitumen linings, 13
air valves, 306307, 307, 310, 311312 Biwater Manual, 4748
booster pumping stations, 196 blackout
low-lift system after pump trip, 191 see also failure
polytropic coefficient, 182 blackout area, 495497
pump blockage, 353 bladder vessels, 219221, 220
pumping failure, 387, 388, 404 blank flange boundaries, 6566, 66
pumping station tripped pumps, 170 blockage
risers, 353 pumps, 351354, 352
Sharjah Pumping Station, 170 booster pumping stations
airwater interface, 215219 flow instabilities, 503
Al Ghouta Irrigation Project, 286291, head conditions, pump start-up, 506
287290 pressure vessels, 193, 194, 195198,
allowable pressures 199200
pipes, 11 booster pumps, 136137, 136, 170
allowable strain borehole configurations, 490
flexible pipes, 447 bottom water level (BWL), 189
allowable surge amplitude boundaries, 6086
plastic pipes, 18 actuated valves, 7177, 7274
Alstom Clasar check valves, 422, 423424 automated control valves, 7577
amplification bifurcation conditions, 8284, 82
networks, 481488 cases, 6769
potential for, 497498 characteristics, 6061, 61
transient pressures, 474498 continuous drawoff, 8486
anticipation valves, 366 definitions, 6061
APEX valves, 295, 296, 307 in-line valves, 7475
applied models isolating valves, 8384
flexible pipes, 465472 large reservoirs, 63
see also modelling non-reflecting, 7882, 79
area blackout, 495, 496497 pipe property changes, 6469, 65, 6769,
asbestos cement, 10 68
attenuation pumping stations, 80, 81, 390394, 391,
closed valve conditions, 100 393
pipelines, 95100, 96 reservoirs, 6264
pump/valve downstream conditions, 100 tanks, 6264
wave heights (zero amplitude), 99 terminal valves, 73
wavefront conditions, 9798 time steps, 7778, 77
automated control valves, 7577 trunk mains, 6971
pressure-reducing valves, 75 types, 6162
pressure-sustaining valves, 7576 BPC see break pressure chamber
surge type, 365366 branches
auxiliary torsional springs, 418 connections, 474476
axial flow pumps, 128 pipes, 6469, 65
see also turbine pumps break pressure chamber (BPC), 508512,
axial springs, 422 508, 512
Aysgarth service reservoir, 193, 194, 195, 196 bridge parapet spillage, 374
buckling pressures
back-flow check valves, 407408 envelope curves, flexible pipes, 466
528
Index
long-term, flexible pipes, 452458 chart usage, 445

short-term, flexible pipes, 463465 closure, 414415, 435445
buffer tanks, 321324, 322 initial closure, 396397
Burnside booster pumping stations, 503, 506 transient pressure reduction, 394396
burst pipes, 485488 computed behaviour, 436, 437
bursting disks, 375, 375 downstream pipework, 328, 329
butterfly valves, 114115, 115 dynamics, 376408
air venting, 336, 338339 free-acting modifications, 414415
as check valves, 421422, 421 head vs. time, 399
BWL (bottom water level), 189 in-line, 406407
bypass check valves, 400401, 401, 407 installation, 377380
bypasses around non-return valves, 186, 187 maximum pressures, 396400
membrane type, 434, 434
case studies multi-pump installations, 388
air valves, 310314, 311313 nozzle type, 422424, 422, 443
booster pumping stations, 193, 194, performance comparisons, 410
195198, 199200 pressure wave reflections, 383385
flywheels, 163167 pumping stations, 377, 378
pressure vessels, 193, 194, 195198, recoil type, 416417, 417, 437, 439, 440
199200 reopening
pumps, 141146, 163167 doors, 382388
sewage pumping pressure vessels, 179, longer term, 385388
180, 181 pressure waves, 383385
sewage rising mains, 345351 responses, 376377, 388, 409410
cast iron, 910 rising mains, 400, 401406
Castlemilk Low pumping station, 503504, rubber flap type, 418420, 419
504, 505 sewage rising mains, 347
cavitation, 33, 102, 104, 105, 397400 shutting, 388390, 399
cement mortar linings, 1314 split disk type, 420421, 420, 441, 442
central tubes, 225226, 226227 submersible pumping stations, 327
centrifugal pumps, 127128, 228 surge behaviour, 388390
Cg function, 463, 464 swing type, 411416, 412, 437438
chambers see tanks tilting disk type, 417418, 418, 437, 439,
characteristic equations, 4159 440
Abbott, 43 vessel connection, 400
application, 4959 circular section deformation, 458462
corrosion, 47 Clachan HEP station, 231232, 231
development, 4144 closed-end boundaries, 6566, 66
fixed grid mesh, 55 closure
fixed wave speeds, 5658, 57 air valves, 308310, 309
integrals, 44, 48 check valves, 388390, 399
large time step, 5556, 56 behaviour prediction, 435445
model outputs, 59 closing moment, 411
natural mesh, 5254, 53 free-acting modifications, 414415
pipe elevation, 4445 initial closure, 396397
pipeline resistance, 4548 nozzle type, 423
system ‘characteristics’, 43 transient pressure reduction, 394396
use, 4952, 50 simplified valve system, 8794
x—t plane, 50, 51, 52 slow-closing valves, 334339
charts coal tar linings, 13
check valves, 445 coast-down velocities, 398
check valves, 186, 187 ColebrookeWhite equation, 46
application alternatives, 400408 collapse
backflow, 407408 vapour cavities, 102, 104
behaviour, 435445 collector pipeline systems, 488490
butterfly valves as, 421422, 421 composite systems, 471, 472
characteristics, 409445 compressible flow theory, 2326, 24, 2526
529
compression wave propagation, 100, 101, Demag DRV-Z valves, 422, 423
103 demand points
computer programs, 3 mains, 85
concrete pipes, 1011 demand-sensing pressure-reducing valves,
condensers, 6 7677
conduits, 2829, 32, 34 densities
configurations modified Proctor density, 456
break pressure chamber, 508 depth of cover
pipelines, 137139 pipes, 463, 464
pumping stations, 137139, 138 descending outfall, 467468, 467468
connections, 236, 400, 474476, 475 differential orifices, 186, 187
conservation of force, 2324, 24 dip-tubes see central tubes
constrained conditions direct pump start, 140, 141146, 142145
flexible pipes, 454458, 463465 envelope curves, 143, 145
constraint factors multi-pump operation, 143146,
parameter, 31 144145
construction discharge
feeder tanks, 271, 273 relief valves, 369
continuous drawoff, 8486 submerged valves, 117118, 117
control valves see valves see also outlets
correction factors disconnecting valves, 286291, 287291
buckling pressures, 455 disks
corrosion equations, 47 bursting disks, 375, 375
costs lift-disk check valves, 424
protection, 200201 split disk valves, 420421, 420, 441, 442
coupling tilting disk valves, 417418, 437, 439,
pumped outfall pipeline, 355, 356 440
covers, 320321, 464 displacement pumps see reciprocating pumps
crevices, 34 distribution
cross-section changes, 476478, 476 free gas, 5859
cross-section of pipeline, 214 distribution systems, 85, 86, 482483
cyclical oscillations, 472473 DN 450 mains, 179, 292, 293, 314316,
cylindrically balanced relief valves, 371, 372, 315317
373 DN 600 mains, 179
DN 700 mains, 292, 314316, 315317
Daer network link, 501507, 502 doors
damage check valves, 382388
air valves, 353 recoil valves, 416, 417
Damascus irrigation project, 286291, reopening, 382388
287290 swing check valves, 413
damped check valves, 415416 double orifice valves (DOVs), 295, 296
damped relief valves, 362 air flow, 299, 300
damper installations, 518519 locations, 297298, 297
dams, 114 downstream head variations
Darcy equation, 46 check valves, 378, 379
definition sketches pump trips, 484
connection analysis, 475 sleeve valves, 429, 430
cross-section change, 476 downstream pipework
deformation analysis, 459 check valves, 328, 329
sleeve valves, 427 downstream pumping failure
definitions, 12 heads, 406
deflections downstream pumping stations, 193,
flexible pipes, 447, 449451, 465, 467 196199
pipes, 12 downstream sleeve volume, 431
deformation, 448, 452, 458462, 459 drainage pumping stations, 292293
Delft Hydraulics, 436 drawoff, 8486
Demag DRV-B valves, 442443 duckbill valves see sleeve valves
530
Index
ductile iron, 11 rigid-column approach, 2123, 22

Duoglide pump case study, 163, 164166, see also characteristic equations
167 external pipeline resistance, 390393
East Main see Glasgow East Main pumping failure

system air valves, 310, 311
effluent outfall systems, 253254 pipe modes, 17
El Yai water transmission system, 335 see also pump trips; pumping failure
elastic buckling, 463465 fast-acting valves, 87, 88
see also buckling pressures feeder tanks, 259278
elastic equations/approach, 56 after pump trip, 271, 274
elevations behavioural aspects, 264267, 264, 266,
pumping stations, 337 276277
wet well/dry well, 346 components, 259261, 260
embedment vs. shape construction site, 271, 273
flexible pipes, 451 elevation, 271, 272
emergency closure, 114115, 115, 119 envelope curves, 268, 269270, 271
energy reservoirs, 159172 examples, 267276
envelope curves heads, 265, 266, 276, 278
air valves, 292, 311 inlets, 271, 273
area blackout, 496497 location, 259261, 260
buckling pressures, 466 mains duplication, 267276
composite system, 472 operation mode, 261264, 262
descending outfall, 467468 pump failure, 268, 269270, 271
direct pump start, 143, 145 pump trips, 271, 274, 277
Duoglide pumpset trip, 164 rising mains, 264
feeder tanks, 268, 269270, 271 volume estimation, 277278
flexible pipes, 466 water levels, 271, 275, 277
gravity mains, 480, 512 filling connections
horizontal pipelines, 465 storage tanks, 284
low-lift systems, 192 filtration plant, 243245, 244246
outfall pipelines, 357 fittings
pressure vessels, 180, 188 pipelines, 7177
pump failure, 147, 148, 150, 348, 368, pressure vessels, 215, 218219
403, 467468, 472 fixed grid characteristic mesh, 5556, 55, 56
pump start-up, 143, 145, 480 fixed wave speeds, 5658, 57
pump trips, 169, 192, 252, 367 flexible pipes, 1112, 446473
pumped outfall pipelines, 357 buckling pressures, 463465
relief valves, 368 cyclical oscillations, 472473
‘rigid’ pipes, 469470 deflection, 449451, 465
rising mains, 468, 469470, 480 deformed shapes, 448, 452, 458462
Sharjah Pumping Station, 169 ductile iron, 11
stiffeners, 468 hydraulic transient analysis, 1112
throttle effects, 188 linings, 1314
Uniglide pumpset trip, 164 long-term buckling pressures, 452458
unsupported pipes, 469 materials, 446448
vacuum disconnecting valves, 290, 291 model applications, 465472
valve closure, 112, 121122, 121 properties, 446448
valve opening, 110 short-term pressures, 463465
water tower pumps, 252 steel pipes, 1112
epoxy linings, 13 strain, 447, 449451
equations, 2126 flow
compressible flow theory, 2326 air valves, 302307, 303
incompressible flow theory applications, air venting, 330
207213 break pressure chambers, 508510
of motion, 201203 Grove regulators, 371
pressure vessels, 200214 multi-pumps, 144146, 145
531
flow (continued ) pump start-up, 480, 481

piston pumps, 150, 151 uniform, 100105
pump initial conditions, 146 grey cast iron, 910
relief valves, 363 Grove regulators, 369371, 370, 371
reversal, 302307, 303 GRP (glass reinforced plastic), 162, 463464
valve opening, 110
flow instabilities, 499519 HQ performance curves, 131
computations, 503 H=N2 vs. Q=N curves, 132
hydropower stations, 513519 H1 valves, 509, 511512
pumping systems, 501507 Haidaria project, 209, 210
remedial measures, 506507, 512, hammering, 1, 361
518519 HDPE see high-density polyethylene
fluids head-loss coefficients, 71, 72
parameter, 2728 heads
flywheels, 159172 3 min valve closure, 111
advantages, 171 air valves, 293, 303, 304305, 308310,
booster pumps, 170 309, 311, 312
case studies, 163167 pump restart, 314, 315
installations, 160, 161, 170171 pump start-up, 331
multi-pumps, 170171 small orifice, 316, 317
pipeline limitations, 162163 air-filled risers, 332, 334
Sharjah Pumping Station, 167170 area blackout, 495
size limitations, 161162 booster pumping stations, 195, 197, 198,
forces at pipe cross-section, 28 506
Fossdale service reservoir, 193, 195 break pressure chamber, 512
free-acting modifications, 414415 Burnside pumping stations, 506
Castlemilk Low, 505
gas, 325358 check valve shutting, 389, 399
distribution, 5859 feeder tanks, 265, 266, 276, 278
envelope curves, 469 gas volume increase, 350
evolution, 339340 gravity mains, 481
fixed-grid computations, 57 horizontal pipelines, 466
influence on parameter, 3238, 37 Lawhead/Silverburn rising main, 166
pipelines, 5859, 340343 long rising mains, 147, 148
pockets, 340343, 341 longer pipelines, 433
rising mains, 345351, 469 losses, 46, 508, 510
sewage mains, 345351 low-lift systems, 190
gas charge/liquid level control, 217218 membrane plants, 246
gas volumes, 210, 342, 350 needle valves, 120121
Geassa tower, UAE, 250 Penn’s Hill, 223
Glasgow East Main pumping system, pipe bursts, 485
501507, 502, 505 pipeline after trip, 312, 313
glass reinforced plastic (GRP) mains, 162, polytropic coefficient, 183
463464 pressure vessels, 184
Glenfield APEX valves, 307 pump start-up, 331, 337, 339, 348
Glenfield check valves, 417, 418, 436, 437, booster pumping stations, 506
439, 440 gravity mains, 481
gradients see hydraulic gradients pump trips, 484
graphical techniques, 177178 comparisons, 349350
gravity flow outlets, 281 composite system, 471
gravity mains, 479481 long main, 367
compression wave propagation, 100, 101, low-lift systems, 190
103 over 8s, 357
envelope curves, 480, 512 pipelines, 312, 313
flow system, 507512 relief valves, 368
heads, 481 Sharjah Pumping Station, 167
hydraulic conditions, 507 sleeve valves, 429, 430, 432
532
Index
pumping failure, 384, 385386 rigid pipes, 911

downstream, 406 secondary objectives, 8
horizontal pipelines, 466 turbine pumps, 132139
upstream, 405 hydrodynamic forces, 356, 358
pumping stations, 268 hydrodynamic moments, 411
after trip, 311, 312 hydropower stations, 513519
Castlemilk Low, 505
failures, 276, 278 impeller type pumps, 127
pump start-up, 337 in-line valves, 7475, 118122
pumps air valves, 406407, 407
blockage, 352 emergency closure, 119
comparisons, 166 isolating, 118119
restart, 358 needle type, 119121
relief valves, 363 inclined pipelines, 234, 235
seal weirs, 250 inclined pressure vessels, 174
seawater intake systems, 248 incompressible flow theory, 207213
Sharjah Pumping Station, 167 complete equations, 207
siphon breakers, 290 transient protection, 201, 203207
sleeve valves, 429, 430, 432, 433 inertia
submerged discharge valves, 517, 519 moment of, 159160, 171, 172
surge tanks, 238 pumpset, 492493
throttled outflow, 344 inertial heads
throttling inflow, 184 valve closure, 93
two-stage valve closure, 116 inflow throttling, 184185, 186, 187188,
valve closure, 104105, 104 189
valve opening, 109 initial gas volume vs. head, 342
vs. flow, 363 inlets
vs. initial gas volume, 342 feeder tanks, 271, 273
vs. ramp slope, 334 inspection
vs. time, 344, 399 air valves, 320
weirs, 250 instabilities
wellfields, 492494 flow, 499519
heat transfer, 175, 176 installation
Henry’s law, 32 air valves, 225229, 226229
high head relief valves, 371375 check valves, 377380
high-density polyethylene (HDPE) pipes, 16 relief valves, 360, 361
high-rise buildings, 378, 379 insulated pressure vessels, 194
horizontal pipelines, 465467, 465, 466 intake systems, 245248, 247
hydraulic conditions, 505, 507 integrals
hydraulic gradients, 317318, 318, 321, 322 equations, 44, 48
in-line check valves, 407 inverted U pipework, 254256
pump trips, 381, 382 irrigation project, 286291, 287290
hydraulic rams, 155156, 156 isolating valves, 8384, 84, 118119
hydraulic transient analysis, 4, 720
buckling pressures, 463465 jet pumps, 156158, 157158
flexible pipes, 1112
flow instabilities, 503 Kauner hydroelectric plant, 187
maximum pressures, 8 Kennedy, -, 435436
minimum pressures, 1819 Kielder project, 209, 210
necessity, 1920 kinetic energy equations, 171, 172
overpressure allowance, 1213 Kirkleatham Lane pumping station, 481488
permitted pressures, 8 Kurdaha project, 209, 210
pipe linings, 1314
pipe materials, 89 large orifices, 310, 314, 315, 344
plastic pipes, 1418 large reservoir boundaries, 6263, 63
pumping station prediction, 380382 Lawhead/Silverburn rising main, 163167,
purpose, 78 164166, 167
533
lift-disk check valves, 424 moment of inertia, 159160, 171, 172

link mains pumps, 505506 moments
Llandinium high-lift pumps, 184, 188 check valves, 411
locations Moody formula, 46
air valves, 297298, 297 motion
feeder tanks, 259261, 260 equation of, 201203
Loch Lomond Scheme, 175, 230, 231 moving ball check valves, 424425, 425,
long horizontal pipelines, 465467 440
long main pump trips, 367 multi-door check valves, 417
long rising mains, 147, 148 multi-pumps, 143146, 144145, 170171,
long-term ovalisation rate, 451452 388
longer-term check valve reopening,
385388 natural characteristic mesh, 5254, 53
losses, 46, 108, 139 needle in-line valves, 119121, 120121
low-level back-flow connection, 282, 283 negative pressures, 259, 260
low-lift systems, 189, 190192, 193 network amplification, 481488
New Aleppo project, 209, 210
mains Neyrtec relief valves, 360361
demand points, 85 non-reflecting boundaries, 7882, 81,
DN type, 179, 292293, 314316, 390393, 391
315317 non-return valves (NRV) see check valves
duplication examples, 267276 nozzle valves, 422424, 422, 423, 443
feeder tanks, 264, 267276 NRV (non-return valves) see check valves
heads, 367 numerical methods
pressure waves, 479481 volume estimation, 178
pump trips, 367
pumping stations, 137, 138 observed behaviour
sliming, 4748 small hydropower stations, 513516
steel gravity type, 107 one-way surge tanks, 241242, 241
suction type, 479481 opening see reopening
valves, 107113, 107112 operating conditions
see also rising mains link mains, 505
mass oscillations, 1 wellfields, 490494, 496497
materials, 321, 446448 operating pumps, 384, 385, 388
maximum expanded gas volume, 207209 see also pump . . .
maximum pressures operating times
hydraulic transient analysis, 8 valves, 122124
plastic pipes, 18 operation
pump trips, 487, 488 air valves, 299, 301
valves, 112, 396400 feeder tanks, 261264, 262
membrane filtration plants, 243245, Grove regulators, 370
244246 valves, 106124, 299, 301
membrane valves, 434, 434 opposing pressure waves, 478479, 478
minimum pressures, 1819, 179 orifices, 186, 187, 316, 317, 344
mixed flow pumps, 128 oscillations, 361, 472473, 499501, 506
see also turbine pumps outfall pipelines, 354358, 467468
modelling outfall systems, 253254
flexible pipe applications, 465472 outflow channels, 514
pressure vessels, 173, 174, 175176 outflow devices, 332333, 343345
pumping stations, 390394 outlets, 279294
submerged discharge valve behaviour, air valves, 292293, 292293
516518 back-flow connection, 282283, 283
see also system curves bellmouths, 279280, 280
models, 4, 6, 59 gravity flow, 281
modifications reservoirs, 114
check valves, 412, 414416 siphon breakers, 284291, 284285,
modified Proctor density, 456 287289
534
Index
ovalisation rates, 451452 pipes

overpressure allowance, 1213 allowable pressures, 11
bursts, 485488, 485
paint linings, 14 cross-sectional forces, 28
Pao La Balsa, Venezuela, 234236, 235 deflections, 12
parabolic approximations, 133 diameters, 11
Parmley, -, 435 elevation, 4445
PE see polyethylene failure modes, 17
peak pressures, 397400 linings, 1314
peak upsurge pressure head, 209212 materials, 89
Penn’s Hill, 222, 223 pressure classes, 15
performance pipework, 328, 329, 338, 347
Castlemilk Low pump, 504 see also pipelines
check valves, 410, 438, 444, 445 piston pumps see reciprocating pumps
Demag DRV-B valves, 442443 plastic pipe analysis, 1418
digitisation pumps, 134 allowable surge amplitude, 18
displacement pumps, 152 maximum pressures, 18
forms, 130 thermoplastics, 1517
Glenfield recoil valves, 439, 440 thermosetting plastic, 1415
HQ curves, 131 pneumatic ejectors, 152155, 153, 154, 154
moving ball valves, 440 polyethylene (PE), 14, 16
pump vs. speed relationships, 328, polypropylene (PP) pipes, 17
329 polytropic coefficient, 182, 183
split disk valves, 441, 442 polytropic relationships, 174176, 299
spring-loaded check valves, 444, 445 porting
swing check valves, 438 valves, 509
tilting disk valves, 439, 440 positioning
turbine pumps, 128132, 134 pressure vessels, 221225
periodic flow, 1 relief valves, 363
permitted pressures, 8 power plant condensers, 6
pipelines, 45, 95105 pressure
parameter, 31 cast iron, 910
attenuation, 95100, 96 check valves, 396400
branches, 6469, 65 contours, pump trips, 487, 488
collector systems, 488490 downstream of check valve, 378, 379
cross-sections, 214 network amplification, 485488
fast-acting valves, 87, 88 pump start/stop operations, 349
fittings, 7177 vs. acoustic wave speed, 37, 37
flywheels, 162163 see also relief valves; transient pressures
free gas distribution, 5859 pressure sensing valves, 365
gas pockets, 340343, 341 pressure surges see hydraulic transient
heads after trip, 312, 313 analysis
horizontal, 465467, 465, 466 pressure vessels, 173229
isolating valves, 8384, 84 air valves, 225229
junctions, 65, 8384, 84 airwater interface, 215219
length, 433 bladder type, 219221, 220
properties, 470472 booster pumping stations, 193, 194,
pumped outfall, 354358 195198, 199200
pumping station boundaries, 390393 central tubes, 225226, 226227
resistance, 4548, 390393 estimation equations, 200214
rising main profile, 222 fittings, 215, 218219
sewage, 3940 gas charge/liquid level control, 217218
sleeve valves, 433 inclined, 174
small hydropower stations, 513 insulated, 194
as surge shafts, 234, 235 locations, 224
water towers, 250252, 251 Loch Lomond (Balmore), 175
see also flexible pipes low-lift systems, 189, 190192, 193
535
pressure vessels (continued ) flywheels, 159172

modelling, 173, 174, 175176 heads, 484
positioning, 221225 comparisons, 349350
rational heat transfer equation, 176 composite system, 471
reversed flow and refilling, 183, 184185, long main, 367
186, 187188 over 8s, 357
Riding Mill, 221225, 221223 relief valves, 368
simplified analysis, 202 sleeve valves, 429, 430, 432
surge suppression, 176, 177 hydraulic gradients, 381, 382
tubes, 225226, 226227 maximum pressure contours, 487, 488
types, 215 pressure vessels, 173214
vertical arrangements, 216 relief valves, 368
volume requirement estimation, 177179 siphon breakers, 290
worst-case conditions, 181183 sleeve valves, 429, 429, 430, 430, 431,
see also air compressors 432
pressure waves velocity, 398, 429, 430, 431
air valves, 301, 302 see also pumping failure
branch connections, 474476 pumped outfall pipeline, 354358, 354
cross-section change, 476478 pumping failure, 146150
meeting of opposing, 478479, 478 air valves, 310
reflections, 239, 383385 air volume, 387, 388, 404
shafts, 236239, 237, 239 envelope curves, 147, 148, 150, 348, 403
speed, deformed circular sections, composite system, 472
458462 descending outfall, 467468
suction mains, 479481 feeder tanks, 268, 269270, 271
valve closure, 89, 91 heads, 385, 386
valve reopening, 383385 downstream, 406
wave fronts, 50 horizontal pipelines, 466
pressure-reducing valves, 75, 7677 upstream, 405
pressure-sustaining valves, 7576 hydraulic gradients, 381
Proctor density, 456 long rising mains, 147, 148
propeller pumps, 128 sewage schemes, 147, 149
properties sleeve valves, 430, 431432
flexible pipes, 446448 three simultaneous trips, 147, 149
pipelines, 470472 velocities, 149, 149, 384, 386, 387, 404
sewage, 3940 VikingJohnson coupling, 356
protection wellfields, 494497
costs, 200201 see also pump trips
flywheels, 159172 pumping stations, 80, 81
pressure vessels, 173214, 177 after trip, heads, 311, 312
rising mains, 479 air valves, 310314, 311313
surge tanks, 236239, 237, 239 arrangement types, 325327, 329
pulsatile flow, 1 centrifugal, 228
pump start-up, 140141 check valves, 377, 378
air-filled risers, 325334 configurations, 137139, 138
booster pumping stations, 506 distribution systems, 482483
gravity mains, 480, 481 drainage, 292293
heads, 331, 337, 339, 348, 481, 506 elevations, 337, 346
pressure variations, 349 failures, 276, 278
pumping stations, 336, 337, 506 flow instabilities, 503504
rising mains, 480 heads, 268, 276, 278
sequenced operation, 493494 after pump start-up, 337
slow valve closure, 334339 after trip, 311, 312
soft start, 333334 hydraulic transients predictions, 380382
pump trips layout, 383, 384
envelope curves, 252, 367, 368 losses, 139
feeder tanks, 271, 274, 277 mains, 137, 138
536
Index
modelling, 390394 refilling pressure vessels, 183, 184185, 186,

network amplification, 481488 187188
non-reflecting boundaries, 80, 81 reflections
relief valves, 366369, 368 check valve reopening, 383385
Riding Mill, 221223, 221223 waves, 101, 104, 383385
sewage, air valves, 310314, 311313 see also non-reflecting boundaries
velocity after pump start-up, 336 reflux valves see check valves
wet well/dry well, 346 regulation
pumps, 125158 gravity flow system, 508510
backwash type, 244 Grove regulators, 369371, 370
blockage, 351354, 352 relief valves, 359375
booster pumps, 136137, 136 behaviour, 363365, 364
case studies, 141146 damped, 362
centrifugal, 127128 discharge, 369
check valve reopening, 384, 385, 386, elements, 360
387 high head, 371375
check valve velocities, 384, 386, 387 installation, 360, 361
definition, 125 positioning, 363
direct start, 140, 141143, 142, 143, pumping stations, 366369, 368
144145 schematics, 364
downstream conditions, 100 sizing, 362363
failure, 146150 types, 359362
flow, 146, 501507 reopening check valve doors, 382388
flywheels, 159172 reservoirs
impeller type, 127 boundaries, 6264
instabilities, 501507 of energy, 159172
jet pumps, 156158, 157158 outlet arrangements, 114
multi-pump operation, 143146, 144145 resistance allowance, 390393, 391
performance, 328, 329 resistance to flow equations, 203207
reciprocating, 150152, 151 resonance, 499500, 501
restart, 314318, 315316, 317, 358 responses
sequenced start, 144, 145 check valves, 376377, 409410
sewage rising mains, 347 multi-pump installations, 388
sleeve valve failure, 432 restarting pumps, 314318, 315316, 317,
speed, 163167, 328, 329, 432 358
stop operations, 349 restricted air outflow devices, 332333
surge effects, 126 reversed flow
system curves, 139140 pressure vessels, 183, 184185, 186,
transfer pumps, 134136, 135 187188
transient behaviour, 125126 reversed velocity
turbine pumps, 126139, 140141 air valves, 306307, 307
types, 125126 operating pumps, 388
wellfields, 492493 RHT see rational heat transfer
see also operating pumps; turbine pumps Riding Mill pumping station, 221223,
221223
Qatar (State), 310 Rifa’a/Hamad Town Blending Station, 189,
190, 191, 192
radial flow pumps, 127128 rigid pipes, 911, 1314, 469470
see also turbine pumps rigid-column approach, 2123, 22
ramp slope vs. head, 334 risers, 325334, 353
rarefaction waves, 101, 104 rising heads, 102, 105
rational heat transfer (RHT) equation, 176 rising mains
reciprocating pumps, 150152, 151, 152 check valves, 400, 401406
recoil valves, 416417, 417, 437, 439, 440 feeder tanks, 264
recorded pipe bursts, 485488 protection of, 479
recorded pressure variations, 349 pump start-up, 480
Redcar distribution area, 482483 rigid pipe, 469470
537
rising mains (continued ) pump trips, 429, 430, 431, 432

sewage case study, 345351, 347 velocities, 429, 430, 431, 433
stiffeners, 468 sliming
uniformly rising, 468470 mains, 4748
unsupported pipes, 469 slow-closing valves, 334339
see also mains small branches, 69, 70
rubber flap valves, 418420, 419 small hydropower stations, 513519, 513
small orifice valves, 316, 317, 344
Scottish Water, 163167 soft-start of pump, 141, 333334
SDVs see submerged discharge valves soil types, 463
seal weirs, 248249, 249, 250 sparg line, 339
seawater intake systems, 245248, 247, speed
248 deformed circular sections, 458462
semi-axial flow pumps, 128 flywheels, 163167
sequenced direct start pumps, 144, 145 pumps, 163167, 328, 329
sequenced pump operation, 493494 sleeve valve failure, 432
service reservoirs (SR), 241243, 241, 242 turbine pumps, 129
sewage vs. efficiency curves, 129
parameter, 3840 see also velocities
air valves, 295, 296, 310314, 311313, spillage, 374
319 spiral casing throttles, 186, 187
case study, 345351 split disk valves, 420421, 420, 441, 442
definition, 3839 spring-loaded check valves, 414415, 418,
pipelines, 3940 420, 422, 444, 445
pneumatic ejectors, 152155, 154 spring-loaded relief valves, 360
pressure vessels, 179, 180, 181 SR (service reservoirs), 241243, 241, 242
properties, 3940 Star/Delta arrangements, 140141
pumping systems, 179, 180, 181, start-up see pump start-up
310314 State of Qatar, 310
pumps, 147, 149, 163, 167170 steady oscillatory flow, 1
rising mains, 154, 154, 345351 steel pipes, 1112, 107
temperature, 39 stiffeners, 467, 468
shafts Stonehouse pumping station, 193, 196199
pressure waves, 236239, 237, 239 stop operations
shape vs. embedment, 451 pumps, 349
Sharjah Pumping Station, 167170, storage tanks, 284286, 284285
292293, 314 stored energy
Shiskine project, 209, 210 flywheels, 159172
short-term buckling pressures, 463465 strain
shutting see closure flexible pipes, 447, 449451
side supports, 452 Strathblane surge tank, 230, 231
Silverburn project, 209, 210 sub-atmospheric pressures, 295, 302
Silverburn/Lawhead case study, 163167, subcontracting, 3
164166, 167 submerged discharge valves (SDVs),
simulation uncertainties, 318319 117118, 117, 513519, 515, 517, 519
simultaneous trips, 147, 149 submersible pumping stations, 325327,
siphon breakers, 284291, 284285, 286, 326327, 329
287289, 290 suction mains pressure, 479481
size sump turbulence, 515
flywheels, 161162 surface areas
relief valves, 362363 tanks, 280282
slanting disk valves see tilting disk valves surge
sleeve valves, 425434, 426 air valves, 295, 298302, 300, 301
definition sketch, 427 anticipation valves, 366
failure, 431432 check valves, 388390
heads, 429, 430, 432 control valves, 365366
longer pipelines, 433 protection, 159172, 173214, 177
538
Index
surge chambers see surge tanks transfer pumps, 134136, 135

surge effect type pumps, 126 transformer starting method, 140141
surge shafts see surge tanks transient pressures
surge tanks, 228, 230258 amplification, 474498
equation applicability, 257258 check valve closure, 394396
filtration plants, 243245, 244246 flexible pipes, 448449, 463465
full-size connections, 236 flow instabilities, 504505
head rise, 238 short-term buckling, 463465
large surge chambers, 233236 sub-atmospheric, 295, 302
one-way, 241242, 241 valves, 122124
parameter estimates, 240 wellfields, 488497
protection extent, 236239, 237, 239 see also hydraulic transient . . .
purpose, 230232 transmission
related structures, 240258 pressure waves
seal weirs, 248249, 249 branch connections, 474476
seawater intake systems, 245248, 247 cross-section change, 476478
service reservoirs, 241243, 241, 242 transmission systems, 335
simple analysis, 232233, 233 trench width
special structures, 253254 pipes, 464
underground, 231232, 231 trips see failure; pump trips
water towers, 249253, 250252 trunk mains, 6971, 70
Swallowglide pump, 167 tubes, 225226, 226227
swing check valves, 411416 tunnels, 31
behaviour prediction, 437 turbine house & outflow channels, 513, 514
modifications, 412, 414416 turbine pumps, 126139
performance curves, 438 HQ performance curves, 131
system curves, 139140, 393394, 393 H=N2 vs. Q=N curves, 132
hydraulic transient analyses, 132139
tanks performance curves, 128132, 134
boundaries, 6264 speed vs. efficiency curves, 129
buffer type, 321324, 322 start, 140141
filling connections, 284 transfer pumps, 134136
of finite area, 280282, 281 turbine runner blades, 34
gravity flow outlets, 281 turbulence, 515
surface areas, 280282 TWL (top water level), 189, 192
see also feeder tanks; surge tanks two-stage valve closure, 113116, 116
tapers, 186, 187
temperature U pipework, 254256
sewage, 39 UAE see United Arab Emirates
terminal valves, 73 ultra-high molecular weight HDPE pipes,
thermoplastics, 1517 16
thermosetting plastics, 1415 uniform gravity mains, 100105
Three Valleys Water North Mymms uniformly rising mains, 468470
Ultrafiltration plant, 243245, Uniglide pump case study, 163, 164166,
244245 167
throttling, 212, 213 United Arab Emirates (UAE), 250, 488,
air valves, 343345 489
inflow, 184185, 186, 187188, 189 unsteady flow, 1
pressure vessels, 184185, 186, 187188, unsupported pipes, 469
189 upstream cavitation
surge shafts, 239 check valves, 397400
tilting disk valves, 417418, 418, 437, 439, upstream head variations
440 break pressure chamber, 512
time steps, 5556, 56, 7778, 77 pump trips, 405, 484
time vs. head, 344, 399 sleeve valves, 429, 430
top water level (TWL), 189, 192 upstream pumping station booster, 194196,
torsional springs, 418, 420, 442 197
539
vacuum breakers, 284, 285 velocities

vacuum disconnecting valves, 286291, air valves, 306307
287291, 290, 291 air-filled risers, 330, 331
valves coast-down, pump trips, 398
actuated, 7177, 7274, 119122 low-lift system after pump trip, 190
automated, 7577, 365366 operating pumps, 384, 388
bifurcation conditions, 8283, 82 pump blockage, 352
boundaries, 7177, 7274, 8283, 82 pump start-up, 336
closure, 8794 pumping failure, 384, 386, 387, 404
0 < t L=, 8990 pumping stations, 336
2L= < t 3L=, 92 sleeve valves, 429, 430, 431, 433
3L= < t 4L=, 9394, 93 tripped pumps, 168, 429, 430, 431
envelope curves, 112, 121122, 121 two-stage valve closure, 116
heads, 104105, 104, 111 vs. initial air volume, 306307, 307
inertial heads, 93 see also speed
instantaneous, 8789 Venezuela
L= < t 2L=, 9092 Pao La Balsa scheme, 234236, 235
max pressure vs. time, 112 venting see air venting
pressure waves, 89, 91 vertical bellmouths, 279280, 280
t ¼ 0, 8789 vertical pressure vessels, 216
times, 122124 Viking SR complex, 242243, 242
two-stage, 113116 VikingJohnson coupling, 355, 356
description, 106 void fractions, 37
downstream attenuation, 100 volume
emergency closure, 114115, 115 buffer tanks, 321324, 322
fast-acting valves, 87, 88 feeder tanks, 277278
flow instabilities, 511512 pressure vessel requirement estimation,
gravity flow system, 509 177179
H1 valves, 509, 511512 volumetric tanks see feeder tanks
head loss characteristics, 510
head-loss coefficients, 71, 72 wafer-type valves, 418, 423424, 442
house & outflow channels, 513, 514 walls
in-line, 7475, 118122 conduits, 34
loss coefficients, 108 Warren Park Pumping Station, 242
opening, 109, 110 water mains see mains
operation, 106124 water towers, 249253, 250252
improvement, 113 waterhammer, 1, 361
times, 122124 wave heights, 99
porting arrangements, 509 wave speed, 30, 37, 37, 458462, 462
slow-closing, 334339 see also pressure waves
small hydropower stations, 513519 wavefronts, 9798
submerged discharge, 117118, 117 Weir Pumps Ltd, 163, 165167
terminal, 73 weirs, 248249, 249
transient pressures, 122124 wellfields, 488497, 489, 491494
vacuum disconnecting type, 286291, Wensleydale Project, 193196
287291 wet well/dry well pumping stations, 325,
see also air valves; butterfly valves; check 346
valves; relief valves
vaporous cavitation, 33, 102, 104 x—t plane characteristics, 50, 51, 52
variable speed starting, 141
variable wave speed , 5455, 55 Yarker Bank reservoir, 193, 195
540

Pressure Transients in Water Engineering

Uploaded by

Copyright:

Available Formats

Pressure Transients in Water Engineering

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Pressure Transients in Water Engineering

Uploaded by

Copyright:

Available Formats

What are some common pipe materials discussed in the document?

What are some common pipe materials discussed in the document?

What factors influence the maximum pressure and allowable amplitude of surge in plastic pipes?

What factors influence the maximum pressure and allowable amplitude of surge in plastic pipes?

Pressure transients in

Distributors for Thomas Telford books are

First published 2008

# Thomas Telford Publishing Ltd 2008

Typeset by Academic þ Technical, Bristol

Chapter 1 Motivation for hydraulic transient analysis 7

Chapter 2 Derivation of basic equations 21

Chapter 4 Characteristic equations 41

Chapter 5 Application of characteristic equations 49

Chapter 7 Valve closure in a simplified system 87

Chapter 8 Actual pipelines 95

Chapter 9 Valve operations 106

Chapter 10 Pumps 125

Chapter 12 Pressure vessels 173

Chapter 14 Surge tanks and related structures 230

Chapter 15 Feeder tanks or volumetric tanks 259

Chapter 16 Discharge conditions 279

Chapter 17 Air valves 295

Chapter 18 Air and gas 325

Chapter 19 Relief valves 359

Chapter 20 Check valve dynamics 376

Chapter 21 Check valve characteristics 409

Chapter 22 Flexible pipe 446

Chapter 24 Flow instabilities 499

Further reading 525

A cross-section area of pipe measured normal to axis (m2 )

circumferential hoop stress in pipe wall (N/m2 )

A number of terms have been associated with the phenomenon of

The distinction between surge or slow transients and waterhammer

in service, leading to an expensive repair job. The failure to appreciate

remove the requirement for surge protection. If understanding of

followed by a study of actual simple pipelines to highlight some of the

1.1 Primary purpose of analysis

pressures. Steady flow is also taken to include static or no-flow

1.2 Secondary objectives

1.3 Permitted pressures

1.4 Maximum pressures

1.5 Pipe materials

1.6 Rigid pipes

1.6.1 Grey cast iron

Nominal Description Maximum working Recommended

80—500 Spigot and socket 10 100 16 160

mWG ¼ metres water gauge

Table 1.2. Test pressures for various classes of cast iron

Nominal internal Description Works proof test pressure

80—500 Spigot and socket centrifugally cast 35 350

Cast iron fails catastrophically.

1.6.2 Asbestos cement

1.6.3 Concrete pipes

Table 1.3. Pressure classes for asbestos cement

Test pressure (bar g) 15 (153 mWG) 20 (204 mWG) 25 (255 mWG)

lengths thus requiring to have a larger number of joints. Unlike other

1.7 Flexible pipes

1.7.1 Ductile iron

1.7.2 Steel pipe

c1 is usually in the range 0:85 c1 1:25