This action might not be possible to undo. Are you sure you want to continue?
.1:
APPLIED HYDRAULIC TRANSIENTS
."!
,',
1
,
,
I
~ }I
APPLIED HYDRAULIC TRANSIENTS
by
."
I'
M. Hanif Chaudhry, Ph.D.
British Columbia Hydro and Power Authority Vancouver, British Columbia, Canada
s
(,
;»"
j:
E:",.:
......
~:fl ;~
Q
.•. ,,.:.....~_ .•. 'J''',., , ,'~•.
==== . __.
PREFACE
Maestro 4~~Gl IN° Copia 4.3£0 ,
no
~======= ....
Van Nostrand Reinhold Company Regional Offices: New York Cincinnati Atlanta Dallas San Francisco Van Nostrand Reinhold Company International Offices: London Toronto Melbourne Copyright © 1979 by Litton Educational Publishing, Inc. Library of Congress Catalog Card Number: 784087 ISBN: 0442215177 All rights reserved. No part of this work covered by the copyright hereon may be reproduced or used in any form, either wholly or in part, or by any meansgraphic, electronic, or mechanical, including photocopying, recording, taping, or information storage and retrieval systemswithout permission of the publisher and the author. Manufactured in the United States of America Published by Van Nostrand Reinhold Company 135 West 50th Street, New York, NY 10020 Published simultaneously in Canada by Van Nostrand Reinhold Ltd. 15 14 13 12 11 10 9 8 7 6 5 4 3 2 1
Library of Congress Cataloging in Publication Data Chaudhry, M. Hanif. Applied hydraulic transients. Includes index. L Hydraulic transients. I. Title. TCI63.C43 620.1'064 784087 ISBN 0442215177
In the last decade, the use of digital computers for analyzing hydraulic transients has increased by leaps and bounds, and the graphical and arithmetical methods for such analyses have been replaced by sophisticated numerical techniques, Not only has this change reduced the amount of laborious computations, but it has resulted in more precise results and has made the analysis of complex systems possible. Applied Hydraulic Transients provides a comprehensive and systematic discussion of hydraulic transients and presents various methods of analyses suitable for digital computer solution. The book is suitable as a reference for practicing engineers and researchers and as a textbook for seniorlevel undergraduate and graduate students. The field of application of the book is very broad and diverse and covers areas such as hydroelectric projects, pumpedstorage schemes, watersupply systems, nuclear power plants, oil pipelines, and industrial piping systems. Each chapter of the book is developed in a systematic manner from first principles, A very strong emphasis is given to the practical applications, and advanced mathematics and unnecessary theoretical details have been avoided as much as possible. However, wherever inclusion of such details was considered necessary from the point of view of researchers, they are presented in such a manner that a practicing engineer can skip them without losing continuity of the text. Several case studies, problems of applied nature, and design criteria are included, which will be helpful to design engineers and will introduce students to the design of reallife projects. Solved examples are given for illustration purposes, extensive lists of uptodate references are included at the end of each chapter for further study, and sample computer programs and flowcharts are presented to familiarize the reader with digital computer applications. Approximate methods and design charts are appended to the text for quick computations during the preliminary design stages. Because of the diverse nature of application, the various chapters have been
v
1. transients in open channels are discussed. Every attempt has been made to identify the source of material. candidate. some of which have not previously been published. and various numerical methods available for their solution are discussed. other parts of the book had to be referred to in order to avoid duplication. Meillquham and G. in addition. Engineer. H. Professor Emeritus E. as well as material drawn from various sources. E.vi Preface 1 written so that they can be read individually. The book presents in a systematic manner a collection of my own contributions. however. Forster. in hydroelectric power plants (Chapter 5).D. reviewed several chapters. for giving me the opportunity and privilege to participate in giving lectures.II ACKNOWLEDGMENTS M. in nuclear power plants (Chapter 6).18. Vancouver. R. Streeter. Hydroelectric Design Division. for granting me permission to include the results of tests conducted on a number of their power plants and a number of design studies for their projects. W. S. Their comments and suggestions for improving the presentation are appreciated. Resonance in pressurized piping systems is discussed in Chapter 8. This modeof presentation will allow practicing engineers to read only those parts of the book that are of their immediate interest. Supervisor. ElFitiany. a brief history of hydraulic transients is presented. E. Ruus. 1. Wylie. Professor of Civil Engineering. a Ph. Vancouver. System Standards Section and 1. I am thankful to him and to the Department of Civil Engineering. S. Senior Engineers. provided encouragement throughout the preparation of the book. These studies were undertaken under the technical and administrative supervision of F. based on parts of this book. Gurney. Patterson.D. H. W. Canada. and various methods for eliminating or alleviating undesirable transients are presented in Chapter 10. Rockwell. any reader preferring to use the English units may do so without much difficulty. was in charge of instrumentation for the prototype tests presented in Sections 3. and 10.11. W. However. SI (Systeme Intemationaley units are used throughout the book.9 and 4. Thanks are extended to my employer. The dynamic and continuity equations for a onedimensional flow in closed conduits are derived in Chapter 2. Dr. Bell. E. University of British Columbia (UBC).8. R. Their excellent publications were a source of inspiration during the preparation of this book. and the details of explicit and implicit finitedifference methods are outlined. and the details of the transfer matrix method are outlined. twophase flows is also presented in Chapter 6. P. E. Engineer. and will allow teachers to select the chapters most relevant to their courses. The analysis of surge tanks using a lumpedsystem approach is presented in Chapter 11. wherever empirical constants are involved or numerical constants are introduced in the derivations. Parmakian. B. My former teacher and research supervisor. E. Arczynski. prepared turbine characteristics vii .8. Supervisor. This has been done in such a manner that only the section referred to may be read and not the whole chapter. D.10. The analysis of transients in homogeneous. Hydraulic Section.8 and 5. Cass. S.and G. British Columbia Hydro and Power Authority (BCHP A). and I. Manager. Hydroelectric Design Division. The details of the method of characteristics are presented in Chapter 3. and to 1. Senior Engineer. In Chapter 12. any oversight in this regard is strictly unintentional. Johnson. W. 5. Martin. Evangelisti. Development Department. A number of design charts and sample computer programs are presented in Appendixes A through D. Pretious and F. M. I would like to express my gratitude to Professors V. C. both of UBC and D. Cass assisted in the studies reported in Sections 4. commonly used terms are defined. The sequence of presentation is as follows: In Chapter 1. L. Sometimes. Transient cavitation and liquid column separation are discussed in Chapter 9. C. and fundamental concepts are introduced. and each reviewed several chapters. Chaudhry. reviewed the entire manuscript and offered many useful suggestions for its improvement. reviewed eight chapters and did the hydraulic transient studies summarized in "Design" on page 324. With these footnotes and the conversion table of Appendix F. Ph. 5. Canada B. Dr. Dirom.7 and devised the gauges for measuring water levels during the tests of Section 12. BCHPA. and in oil pipelines (Chapter 7). _. Manager. Assistant Manager. prepared Figs. their corresponding values in the English units are given in the footnotes. to the graduate students. The next four chapters are problemoriented and discuss transients in pumping systems (Chapter 4).C.
and to the following members of the department: D.. I am grateful to the Department of Water Resources.. 1 would like to extend my thanks to F. Scotland."_.4 2. for several discussions during studies of Sections 3. E. Engineering.. H. and W.7 2.S Assumptions Dynamic Equation Continuity Equation General Remarks on Dynamic and Continuity Methods for Solving Continuity and Dynamic Velocity of Water hammer Waves Case Study Summary Problems References Additional References ix Equations Equations .S. Wong for preparing the figures and illustrations. for many hours that should have been spent with them but were required in the preparation of this book. and 10.5 2. the Water Power and Dam Construction.. 1. CONTENTS PREFACE ACKNOWLEDGMENTS INTRODUCTION v vii 1.. W.:. Reid for typing the manuscript and to L. conducted the hydraulic model studies of Section 12. for granting permission to include material from their publications. and Glenfield & Kennedy Ltd.. Smith. Gibbs. Johnson. Senior Engineer.4 1.3 1.."."' . State of California. Hydraulic Division.7.i:r. L. A. and S.7 I . VicePresident.S. Kingston. Ching and D. Queen's University. Manager. Chauhan..II viii Acknowledgments ~. Myles and P. I am thankful to her and to our son.7. Dr.. Klohn Leonoff Consultants Ltd. Shamim.2 1. Atomic Energy of Canada. :~. Alam... did the programming for the case study of Section 12.6 2. Ontario. Well. H. I am thankful to Westinghouse Electric Corporation for granting permission to include material from their reports in Section 6. and the L'Energia Elettrica for allowing inclusion of material from their copyrighted publications. Associate Mechanical Engineer.S.: •.. and 1. Geets. Lasalle Hydraulic Laboratory. Engineers. H. Wood. Kao. Head.. Whitehead. F._ . Senior Mechanical Engineer. VicePresident. C.'~. L.. G.1 1. 10.9.. K.3 2. Portfors. F. and Engineers L..j'fC' ••. and 10..S and to J. E. j !' d l_: :. . British Hydromechanic Research Association..15. Howard. S. for providing photographs for Section 1. T. E.... International Association for Hydraulic Research. to Professor F.". and L... Head.. 5.. for the studies of Sections 5. Senior Engineer.5.. D. O. Assistant Mechanical Engineer. W.7 and to Ebasco for granting permission to include these photographs in the book._. for providing the test data of Section 4.15.. Westinghouse.1 2. My wife. M.... Asif. Gurney assisted in the proofreading.. Parkinson. . M.8 Definitions Historical Background Pressure Changes Caused by an Instantaneous Velocity Wave Propagation and Reflections in a Single Pipeline Classification of Hydraulic Transients Causes of Transien ts System Design and Operation Summary Problems References OF UNSTEADY CONDUITS FLOW THROUGH Change I 2 7 II 16 IS 19 21 22 22 ... Cass. American Society of Mechanical Engineers. _ .. Hancox. Chaudhry 2. prepared Fig. Thermalhydraulics Research Branch. 5.. helped in numerous ways in completing the manuscript.•• ~. A...2 2.. BOP Systems Design. ) 2 EQUA nONS CLOSED 27 27 27 30 33 34 34 39 40 41 41 43 M. Thanks are also due to the American Society of Civil Engineers. to C.. who enthusiastically furnished a number of relevant reports. Vandenburg.i .. and to Dr.6 1. . New York.5. Ebasco Services Inc.:...5 1.
1 4.x Contents 3 METHOD OF CHARACTERISTICS 3.5 Introduction Schematic of a Hydroelectric Power Plant Upstream and Downstream Conduits Simulation of Turbine Hydraulic Turbine Governors Actuator Dashpot Permanent Drop Distributing Valve Gate Servomotor Computational Procedure 5.3 Introduction Characteristi~ Equations Boundary Conditions ConstantHead Reservoir at Upstream End ConstantHead Reservoir at Downstream End Dead End at Downstream End Valve at Downstream End Orifice at Lower End Series Junction Branching Junction Centrifugal Pump at Upstream End Francis Turbine at Downstream End Stability and Convergence Conditions Selection of Time Incremen t for a Complex Piping System Combined ImplicitCharacteristic Method Analysis of a Piping System Case Study Summary Problems References Additional References 44 44 44 50 51 52 53 54 55 55 57 58 58 58 59 61 63 65 70 70 71 73 74 74 74 75 77 79 81 82 83 83 86 86 88 92 94 95 95 Emergency Catastrophic Verification of Mathematical Model 4.4 Introduction Transient Conditions Caused by Various Pump Operations Mathematical Representation of a Pump Boundary Conditions for Pump Failure Equations of Conditions Imposed by Pump Differential Equation of Rotating Masses Characteristic Equation for Discharge Pipe Continuity Equation Solution of Governing Equations Boundary Conditions for Special Cases Parallel Pumps Series Pumps Example 4.8 Prototype Tests Prototype Data Comparison of Computed and Measured Results 5.6 5.7 3.2 5.1 0 Generator Inertia 109 109 109 110 111 118 122 122 122 122 123 124 124 126 126 127 129 130 130 133 133 133 4 TRANSIENTS CAUSED BY CENTRIFUGAL PUMPS 4.2 4.7 4.6 3.9 I 5 HYDRAULIC TRANSIENTS IN HYDROELECTRIC POWER PLANTS 5.8 3.8 I .1 3.7 Causes of Transien ts Verification of Mathematical Model 5.4 5.3 5.2 3.5 3.9 Design Criteria for Penstocks Normal Emergency Catastrophic 5.4 3.1 5.11 Summary Problems References Additional References Contents xi 96 96 96 97 97 97 98 99 100 100 102 103 104 104 106 107 3.6 4.3 4.5 4.9 Plant Data Tests and Instrumentation Mathematical Model Comparison of Computed and Measured Results 4.10 Case Study WaterSupply System Analysis Results Discussion 4.1 Pump StartUp Design Criteria for Pipelines Normal 4.
I ' ' 6 HYDRAULIC TRANSIENTS 6.13 Summary Problems References Additional References IN NUCLEAR POWER PLANTS 135 135 137 139 140 142 145 145 147 153 153 155 156 158 Contents :t xiii 6.7 6.4 8.3 7.1 6.3 8.8 Introduction Terminology General Types of Reactors Emergency Core Cooling Systems Causes of Transients Methods of Analysis General Remarks Formulation of Mathematical Models Numerical Solution Boundary Conditions Condenser Entrapped Air Pipe Rupture and Failure of Rupture Discs LossofCoolant Accident Implicit FiniteDifference Method for Analyzing TwoPhase Transient Flows General Governing Equations Conversion of Governing Equations into Characteristic Form Formulation of Algebraic Equations Computational Procedure Verification Case Study Configuration of Feedwater Line Description of Incident Possible Causes of Shock 158 158 158 159 160 161 162 162 163 163 165 165 168 169 169 170 170 170 172 174 177 177 178 178 179 182 .8 Introduction Development of Resonating Conditions Forced and SelfExcited Oscillations Forcing Functions SelfExcited Oscillations Methods of Analysis Analysis in Time DomainMethod of Characteristics Analysis in Frequency Domain Terminology SteadyOscillatory Flow Instantaneous and Mean Discharge and Pressure Head Theoretical Period Resonant Frequency State Vectors and Transfer Matrices Block Diagrams Derivation of Transfer Matrices Field Matrices Point Matrices Frequency Response Fluctuating Pressure Head Fluctuating Discharge 201 201 206 206 206 208 209 209 210 210 211 211 212 212 215 216 216 226 235 237 238 .5 'i~ '.1 Transient Speed Curve Optimum Values of Governor Parameters Example 5.12 Case Study 5.3 6.6 8.6 Introduction Definitions Causes of Transien ts Method of Analysis Design Considerations General Remarks Control and Surge Protective Devices Summary Problems References Additional References PIPING SYSTEMS 190 191 194 195 197 197 198 199 199 199 200 201 8 RESONANCE IN PRESSURIZED 8.5 8.2 7. .5 6.11 Governing Stability General Remarks Differential Equations of the System Criteria for Stability Example 5.7 8.2 8. 7.4 7.2 6.1 8.6 6.xii Contents 5.2 5.1 7..9 Summary Problems References Additional References IN LONG OIL PIPELINES )86 186 187 189 190 ' I 7 TRANSIENTS 7.4 6.
8 Introduction Available Devices and Methods for Controlling Transients Surge Tanks Description Boundary Conditions Air Chambers Description Boundary Conditions Valves Description Boundary Conditions Optimal Control of Transient Flows Case Study Design Mathematical Model Results Summary Problems References Additional References 11 SURGE TANKS 11.5 10.8 Numerical Solution Wave Equations Column Separation Cas Release Results 9.1 9.5 i i ~ 11.7 9 TRANSIENT Introduction General Remarks Causes of Reduction of Pressure to Vapor Pressure 904 Energy Dissipation in Cavitating Flows 9.2 11.2 9.1 11.14 267 269 269 271 272 274 274 274 276 278 279 281 283 284 284 285 285 286 287 287 290 291 292 292 293 293 294 295 296 IDA 10.6 10.10 8.2 10.8 Introduction Types of Surge Tanks Derivation of Equations Dynamic Equation Continuity Equation Available Methods for Solving Dynamic and Continuity Equations Period and Amplitude of Oscillations of a Frictionless System Stability Normalization of Equations PhasePlane Method 332 333 334 335 337 337 338 340 342 342 .7 11.7 Derivation of Equations Assumptions Continuity Equation Momentum Equation Cavitating Flow Column Separation 9.9 8. 8.4 11.10 Case Study Project Details Field Tests Mathematical Model Comparison of Compu ted and Measured Results 9.9 Design Considerations 9.3 J.6 Analysis of Cavitating Flows with Column Separation 9.3 10.3 11.6 11.xiv Contents Oscillating Valve Procedure for Determining the Frequency Response Pressure and Discharge Variation along a Pipeline Location of Pressure Nodes and Antinodes Series System Determination of Resonant Frequencies Verification of Transfer Matrix Method Experimental Results Method of Characteristics Impedance Method Energy Concepts Studies on Pipelines with Variable Characteristics Summary Problems References Additional References CAVITATION AND COLUMN SEPARATION Contents xv 8.11 Summary Problems References Additional References 298 298 299 301 TRANSIENTS 10 METHODS FOR CONTROLLING 302 302 303 303 303 304 306 306 308 312 312 317 323 324 324 325 326 328 328 329 330 332 10.13 8.12 240 241 246 248 249 250 256 257 257 260 i64 9.5 Wave Velocity in a CasLiquid Mixture 9.1 10.11 8.
7 12.. : 12 TRANSIENT FLOWS IN OPEN CHANNELS 12.16 12.xvi Contents I .6 12.10 12.11 11.15 i.14 1l.11 .5 Introduction Definitions Causes of Transients Wave Height and Celerity Rectangular Channel Derivation of Equations Continuity Equation Dynamic Equation 382 382 384 384 387 389 390 392 Methods of Solution 394 Method of Characteristics 395 Explicit FiniteDifference Method 397 Diffusive Scheme 398 Formulation of Algebraic Equations 398 Boundary Conditions 399 404 Stability Conditions 404 Computational Procedure 405 Initial Conditions Verification of Explicit FiniteDifference Method408 Diffusive Scheme 408 Mathematical Model 408 Prototype Tests 411 Comparison of Computed and Measured Results 414 Implicit FiniteDifference Method 414 Description 415 Available Implicit Schemes 415 Strelkoff's Implicit Scheme 420 Systems Having Branch and Parallel Channels 422 Stability Conditions Comparison of Explicit and Implicit FiniteDifference Methods 422 423 Special Topics 423 Dambreak 424 Tidal Oscillations 425 Secondary Oscillations or Favre's Waves 426 FreeSurfacePressurized Flows 431 LandslideGenerated Waves 433 Case Study 433 Project Details 433 Mathematical Model 440 Results 440 Operating Guidelines 442 Summary 442 Problems 444 References 'tl APPENDICES Appendix A: Formulas and Design Charts for Preliminary Analysis Appendix B: Computer Program for Analyzing Transients Caused by Opening or Closing a Valve I '1 f iiii 449 469 .8 12.3 12.iJ Contents xvii 11. E 344 344 348 352 356 360 360 360 361 362 362 363 364 365 365 365 365 368 368 369 370 370 371 372 372 373 375 376 376 379 382 .10 1l.9 11. 12.2 12.1 12.9 12.l5 Analysis of Different Cases of Flow Demand ConstantFlow ConstantGate Opening Constant Power Constant Power Combined with ConstantGate Opening Conclusions Orifice Tank Description Derivation of Dynamic Equation Differential Surge Tank Description Derivation of Equations Multiple Surge Tanks Design Considerations Necessity of a Tank Location Size Case Study Project Details Preliminary Investigations Selection of Method of Analysis Program of Investigations Selection of Range of Various Variables Derivation of Equations Analog Simulation Results Selection of Tank Size Summary Problems References 12.13 11.12 12.r !l .14 ~ ~ .j 12.12 11.13 12.4 12.
. .....~""._... '" " .•.<'" ...''''''7''''''''''''':~ xviii Contents Appendix C: Computer Program for Analyzing Transients Caused by Power Failure to Centrifugal Pumps Appendix D: Computer Program for Determining Frequency Response of a Series System Appendix E: Pump Characteristic Data Appendix F: SI and English Units and Conversion Factors I i t'~..... F~..."."~.~ ._" _ 474 48 I 484 487 489 493 AUTHOR INDEX SUBJECT INDEX APPLIED HYDRAULIC TRANSIENTS .. I _. \:.~_""""_'''.~.:. ""'.lr::".. \1 " . I........... . .. ."~..
such as pressure. the flow is termed unsteady. If the flow conditions are varying with time and if they repeat after a fixed time interval. A description of the propagation and reflection of waves produced by closing a valve at the downstream end of a single pipeline is presented. at a point do not change with time. When referring to the steady or unsteady turbulent flows herein. the flow is called nonuniform. SteadyOscillatory or Periodic Flow. a number of commonly used terms are defined. these flows are considered as steady if the temporal mean conditions do not change with time. when the flow conditions are changed from one steadystate condition to another steady state. less common terms are defined in the text wherever they appear for the first time. the flow is called steady . velocity. If the flow conditions. by considering temporal mean values over a short period.1 DEFINITIONS Terms commonly used are defined in this section. The intermediatestage flow. then the flow is said to be steady. Strictly speaking. turbulent flows are always unsteady since the conditions at a point are changing continuously. If the velocity' is constant with respect to distance at any given time. TransientSlate or Transient Flow. 1. and a brief history of the development of the knowledge of hydraulic transients is presented. the flow is called uniform flow. whereas if the velocity varies with distance. we will use the temporal mean conditions. is called transientstate flow or transient flow. If the conditions change with time. The basic waterhammer equations for the change in pressure caused by an instantaneous change in flow velocity are then derived. Uniform and Nonuniform Flow. This is followed by a discussion of the classification and causes of hydraulic transients.CHAPTER 1 ~I INTRODUCTION In this chapter. and discharge. However. Steady and Unsteady Flow.
the flow is steady when the conditions are constant with respect to time (i.lb) __. having frequency wf. Rternanrr'" developed and applied a threedimensional equation of motion and its Simplified onedimensional fonn to such . Now let us consider another situation. He reasoned that the theoretical velocity would increase by about 20 percent if the adiabatic conditions were used instead of the isothermal conditions. 1. Monge developed a graphical method for integrating the partial differential equations" and introduced the term method of characteristics.4 mls as compared to their experimental value of 348 m/s. the propagation of waves in shallow water. in cycles/s expressed in rad/s is called Column Separation. 1. Principiai Introduction 3 oscillatory flow and the referred to as the period. Both Newton and Lagrange obtained theoretically the velocity of sound in air as 298. the flow conditions in the pipeline will become periodic too. Waterhammer. for t < to and t > t I). none of these problems could be solved rigorously until the development of the theories of elasticity and calculus. Newton derived an incorrect expression for the celerity of water waves in a canal as rrVlJi. this wave is dissipated as it travels in the pipeline. the flow through the valve is instantly reduced to zero.2 HISTORICAL BACKGROUND* the solution of partial differential equations. However. He correctly attributed this difference to the elasticity of pipe walls. to obtain a solution for the flow of blood through arteries." Lagrange analyzed" the flow of compressible and incompressible fluids. In the past. x = the equilibrium position of a particle. By comparing the oscillations of a liquid in a Utube to that of a pendulum. in which F and i= the travelling waves. pressure rises at the valve. Laplace" pointed out the reasons for the difference between the theoretical and measured values of the velocity of sound in air. respectively. and the intermediate flow (i. at time t 1 the pressure in the entire pipeline becomes equal to the reservoir head. the water is flowing with velocity Va' and at time. In 1789. and steamhammer referred to the pressure fluctuations caused by a flow change depending upon the fluid involved. Based upon the definitions given previously. however. in which d = canal depth. bend losses. Newton presented. whereas Newton explained that the theoretical velocity was incorrect and that this discrepancy was due to spacing of the solid particles of air and the presence of vapors in air. and the propagation of pressure waves in pipes. and finallylet us say. Let the valve be opened and closed periodically at a frequency. If the pressure in a closed conduit drops below the vapor pressure of a liquid. friction losses. The following discussion will be helpful in clarifying the preceding definitions. the term hydraulic transient is used more frequently. Euler also tried.The study of hydraulic transients began with the investigation of the propagation of sound waves in air. Euler" developed a detailed theory of the propagation of elastic waves and derived the following partial differential equation for wave propagation: =a2 a2y a t2 a2y ax2 (l. He also derived a correct expression for the celerity of waves in a canal as c = y'gd. Helmholtz appears to be the first to point out that the velocity of pressure waves in water contained in a pipe was less than that in unconfined water. readers interested in the history of hydraulics should see Ref. About 1808. In 1869. y = the particle displacement. This flow is called steadyoscillatory flow. and the flow of blood in arteries. Young? investigated the flow of bloodstreams.Ia) in which a2 = gh. Frequency circular frequency and is usually designated by w. and flow is completely stopped.2 Applied Hydraulic Transients time interval at which conditions are repeating is. t = to' the valve is suddenly closed.. If T is the period in seconds.. Due to friction losses. He explained that the relationships derived by Newton and Lagrange were based on Boyle's law and that this law was not valid under varying pressures since the air temperature did not remain constant. As a result of the valve closure. but failed. oscillations.2 is based on Ref.at) (l. and *Most of the material presented in Section 1. and because of the conversion of the kinetic energy into elastic energy. Let us assume that the downstream valve of the pipeline (see Fig. and h = height of the air column. Nowadays. . I. He also developed a general solution of this equation as y '" F(x + at) + f(x .e. oilhammer. then the frequency of and in rad/s is IITand 2rr1T. then cavities are formed in the liquid and the liquid column may separate. and a pressure wave travels in the upstream direction. in his the results of his investigations on the propagation of sound waves in air and on the propagation of water waves in canals. he developed the concept of velocity potential. where L = the wavelength and g = acceleration due to gravity. This wave is reflected from the reservoir and travels back and forth between the valve and the reservoir.e. to ~ t ~ tl) when the conditions are changing from the initial steady state to the final steady state is transient flow. f. wf' After a number of cycles. Lagrange erroneously attributed this difference to experimental error. 2. For this purpose.1 a) is fully open. terms such as waterhammer.
Westorr'" and Carpenter.5 mm.30 demonstrated that doubling of pressure head is not possible unless Ho > aVo/g· Constantinescu " described a mechanism to transmit mechanical energy by using the waterhammer waves. In 1897. T". Allievi"? also studied the rhythmic movemen t of a valve and proved that the pressure cannot exceed twice the static head. proportional to the elasticity of the tube. He studied the effects of air chambers. Resal'? developed the continuity and dynamic equations and a secondorder wave equation. Gromeka included the friction losses!" in the analysis of waterhammer for the first time. Allievi is still considered to b~ the originator of the basic waterhammer theory. 24. Braurr" claimed priority over Allievi. Introduction . Frizell22 presented an analysis of waterhammer based on studies undertaken while acting as a consulting engineer for the Ogden hydroelectric development in Utah. p = onehalf of the ratio of the kinetic energy of the fluid to the potential energy stored in the fluid and the pipe walls at pressure head Ho.2) aTe 2/. interested readers should consult Refs. was actually introduced by Braun. He showed that the term V(o V/ox) in the dynamic equation was not important as compared to the other terms and could be dropped. surge tanks..?? In a later publication. 4 Applied Hydraulic Transients ~~ "1"" . He used Marey's experimental results to verify his analytical studies. He also investigated the effects of the variation of closing rates of a valve and found that the pressure rise was a maximum for closing times. He assumed. He also developed the relationship between the velocity reduction and the resulting pressure rise by using two methods: the conservation of energy and the continuity condition. Joukowski and Allievi. Frizell developed expressions for the velocity of waterhammer waves and for the pressure rise due to instantaneous reduction of the flow.* Joukowski's and AlIievi's theories were mainly used in the first two decades of the 20th century. he published his classic reporr'? on the bas~c U:eory of waterhammer." both American engineers. in which. 101. Unfortunately. in which L = length of the pipeline and a = wave velocity.5 mm. earlier investigators had considered only one of the two at a time. Korteweg'S was the first to determine the wave velocity considering the elasticity of both the pipe wall and the fluid. 152. that the liquid was incompressible and that the friction losses were directly proportional to the flow velocity. however. However. and it appears that the socalled AlIievi's constant o.5 fields as vibrating rods and sound waves. AlJievi obtained an expression for the pressure rise at the valve and presented charts for the pressure rise and drop caused by a uniformly closing or opening valve. He discussed the propagation of a pressure wave along the pipe and the reflection of the pressure waves from the open end of a branch.. 50 mm. i i: _. .' Zl. Joukowski conducted extensive experiments in Moscow on pipes with the following dimensions (expressed in length and diameter. He developed a formula for the wave velocity. Vo = steadystate velocity. waterhammer wave velocity._ . . He also developed the dynamic and continuity equations that are the bases of our studies. Marey 12 conducted extensive series of tests to determine the velocity of pressure waves in water and in mercury and concluded that the wave velocity was: I. This power plant had a 9449mlong penstock. three times greater in mercury than in water 3. Frizell's work has not been appreciated as much as that of his contemporaries. taking into consideration the elasticity of both the water and the pipe walls. He introduced two dimensionless parameters.27. independent of the amplitude of the pressure waves 2. recent investigations by Anderson!" have shown that actually Menabrea'" was the first to study this problem. Michaud " studied the problem of waterhammer. and 305 m. For the valveclosure time. In World *For details of Allievi's work. =Iength of the pipeline. and spring safety valves on waterhammer pressures. 305 m. and successive waves on speed regulation. He also discussed the effects of branch lines. Braun2s. Although Wood I lists Michaud 16 as the first to deal with the problem of waterhammer..]«.26 presented equations similar to those presented by Allievi in his second publication. In 1877. Te. a =. Lord Rayleigh published his book on the theory of sound. and the design and use of air chambers and safety valves. I.!" which summarized the earlier studies and his own research. wave reflections.24 The dynamic equation that he derived was more accurate than that of Korteweg. and 29. Camichel et al. p=  2gHo avo} 8 = (1._ . He stated that the wave velocity would be the same as that of sound in unconfined water if the modulus of elasticity of the pipe walls was infinite. respectively): 7620 m. and () = the valveclosure characteristics. However. conducted a number of experiments to develop a theoretical relationship between the velocity reduction in a pipe and the corresponding pressure rise. Allievi developed the general theory of waterhammer from first principles and published it in 1902. neither one succeeded because their pipelines were short. Te :::valveclosure time. Based on his experimental and theoretical studies. Weber" studied the flow of an incompressible fluid in an elastic pipe and conducted experiments to determine the velocity of pressure waves.
These and important contributions of others are listed in Chapter 3. The unsteady flow situation of Fig. several papers were presented on the analysis of waterhammer in penstocks and in discharge pipelines. respectively. the moving wave front appears stationary (Fig.68 Jaeger. Lais7 used it in his doctoral dissertation.49 From 1940 to 1960. At this symposium. Waterhammer pressures.e.:lV. On the theory of surge tanks. V. nonlinear friction losses in the analysis.50 Jaeger.l'' and DUC.74 and Marris.s2 numerous papers were published on the analysis of waterhammer."! Evangelisti. l."? Rateau. on the control volume.lb)."? Richard. sequence. changes in the turbine efficiency at various gate openings. the fluid density Po is changed to Po + .53b was the first to present procedures for determining a valveclosure Let us consider the piping system of Fig. Based on Joukowski's theory. papers were presented on the analysis of air chambers and valves."! "I 1. This is equivalent to assuming that an observer is traveling in the upstream direction with velocity a. on the inclusion of complete pump characteristics.. Let us consider distance. the pressure at the valve becomes Po + Sp. x.6 Applied Hydraulic Transients Introduction 7 War I.I? Paynter. early European contributions were made by Leaute. He also studied resonance caused by periodic valve movements and pressure drop due to gradual opening of valves and gates. Later on. let us assume that the pipe is rigid. Thorna'" was the first to show that the surge tank of a governed hydraulic turbine would be stable only if the crosssectional area of the surge tank were more than a certain minimum value. L6wy36 independently developed and presented an identical graphical method in 1928. and the uniform and nonuniform gate movements were considered in the analysis. for the first time. British fighter planes were equipped with the Constantinescu gear for firing the machine guns. Gibson32 presented a paper that included. and a pressure wave of magnitude Ap travels in the upstream direction. Cabelka and Franc. and on the comparison of the computed and measured results. Because of their large number. called optimum valve closure. In his discussion of Strowger and Kerr's analysis. Instead. Another symposium'? on waterhammer was held in 1937 at the annual meeting of the American Society of Mechanical Engineers.64 and Frank and Schuller'" summarized the earlier investigations and their own research. Other papers on watercolumn separation were published by Lupton. in which a fluid is flowing with velocity Vo' and the initial pressure upstream of the valve is Po. to simplify the derivation. and later Rich4s used Laplace transforms for the analysis of waterhammer in pipelines. and Schnyder " was the first to include the friction losses in the graphical analysis. 1. they are not listed herein. Other coutributors to the theory of surge tanks are Escande. RUUS53a.3 PRESSURE VELOCITY CHANGES CHANGE CAUSED BY AN INSTANTANEOUS Angus" outlined basic theory and some applications of the graphical method including "lumped" friction losses. the velocity changes to Vo + . Johnson"? invented the differential surge tank to develop accelerating or decelerating heads rapidly. and velocity. velocity a in the downstream direction. so that the maximum pressure remained within the prescribed limits.1.f? Binnie. WOOd35 introduced the graphical method for waterhammer analysis. At a syrnposium"" sponsored jointly by the American Society of Civil Engineers and the American Society of Mechanical Engineers in 1933 in Chicago. He considered the friction losses in his analysis by including the friction terms in the basic partial differential equations.73. Streeter published numerous papers on the method of characteristics as well as a texr" on hydraulic transients. 1.:lp. its diameter does not change due to pressure changes. and. Angus'" presented in 1938 the analysis of compound and branching pipelines and watercolumn separation. Later on. and the inflow and outflow velocities from the control volume are (Vo + a) and (Vo + AV + a). in addition to books by Rich. Schnyder '? included complete pump characteristics in his analysis of waterhammer in pipelines connected to centrifugal pumps.1a is converted into a steady condition by superimposing.70 Gardel. Bergeron:" extended the graphical method to determine the conditions at the intermediate sections of a pipeline.54 and Streeter" independently developed the concept and the latter extended and computerized it for complex piping systems. He also invented an apparatus" to measure the turbine discharge using the pressuretime history following a load rejection.63 Calame and Gaden.5l." and Parmakian. Strowger and Kerr presented'" a stepbystep computational procedure to determine the speed changes of a hydraulic turbine caused by load changes. and his joint paper with Streeter'" was the pioneer publication that made this method and the use of computers for the analysis of transients popular.v' Pnisil. Gray56 introduced the method of characteristics for a computeroriented waterhammer analysis. and Bergeron'f presented a paper describing the theory of plane elastic waves in various media. now commonly known as the Thoma area.62 and Vogt. important contributions are discussed and listed in the following chapters. If the valve setting is changed instantaneously at time t = 0. Wood'" used Heaviside's Operational Calculus. Let us designate the velocity of propagation of the pressure wave (commonly called waterhammer wave velocity) by a. as positive in the downstream . To this observer. i. By linearizing the friction term.
(Po + Ap)A. i. It was assumed previously that the fluid density changes to Po + Ap as a result of the change in pressure. Proceeding similarly.10) . 1.]AH Vollltl +a)] + a)A (1. For the control volume of Fig.3) Conlrol IIOIUmtl) Neglecting friction.5 may be neglected.. MOiling WOlltlfronl Introduction direction.8) Vo+a Vtllocily Otlnsily Pressurtl 1 \ The negative sign on the right hand side of Eq. 1. iSPoA . 1. Vo in Eq.9) ( b) Unsteady flow converted to steady flow by superimposino velocity a (Ap og a Figure l.8 was derived for the case of velocity changes occurring at the downstream end of a pipe and for the wave front moving in the upstream direction. Hence. Then the rate of change of momentum in the positive xdirection 7Po(Vo+a)A = Po(Vo [(Vo+~V+a)(V ~V o 9 ~~t'INtlstlrlloir .9. since p = pgH. 1.4 that ( 1." "i Vo _ (1. the pressure increases for an increase in velocity and the pressure decreases with a decrease in velocity.3 and 1.5) We will see in Chapter 2 that in most of the transient conditions in metal or concrete pipes or in the rock tunnels. in which H is the piezometric head. Eq.6) (1.. 1. ~H is positive) for a reduction in velocity (i.~V g Vo+a Po Po Vo+ AV+a Po+AP Po+AP (1. Hence. F. = Note that there is no negative sign on the righthand side of Eq.e.. for negative ~V) and vice versa. 1.V7 g ~~~~~Ntlstlrlloir Control IIOIUmtl""\ 'i (I. acting on the fluid in the control volume in the positive xdirection. if the velocity was changed at the upstream end and the wave was moving in the downstream direction. This shows that in this case.e. a (approximately 1000 m/s) is much greater than Vo «10 m/s). 1. the resultant force.5 may be written as Ap=poaAV[ Vtllocily Otlnsily Prtlssurtl Vo Po Po Vo + AV Po + Ap Po + Ap (a) Unsteady flow (I.e.4) " According to Newton's second law of motion. Rate of mass inflow = PoA (Vo + a) (l. then ~H= E. Also note that Eq.1b. Also. it can be proved that.7) or ~""'WI'~I w=~b. the time rate of change of momentum is equal to the net force. F= ApA . Pressure rise in a pipeline due to instantaneous reduction of velocity H}. it follows from Eqs.l.8 Applied Hydraulic Transients Initiol sltlody stat« hydroulic grodtl lin.8 indicates that the pressure increases (i.
Example 1. the flow velocity at the valve is reduced to zero. 0<t~L/a(Fig.own by dotted lines). Determine the pressure rise if a steady flow of 0. 6.196 m2 1T Vo = Qo/A = 0.81 (2.l. and all the kinetic energy has been converted mto elastic energy. Po Since (Vo ~P (1. If the system IS assumed fri~tio~les~.flow velocity is zero.4 m3/s is instantaneously stopped at the downstream end by closing the valve. Solution: A ~.V+a).2) as follows: following the valve closure can be divided into four apo ~a=  LV /K '.1. If a is the velocity of the waterhammer waves l.Smdiameter pipe conveying oil from a reservoir to a valve. Therefore.17) Let us consider the piping system shown in Fig. therefore..04) = 268.4 WAVE PROPAGATION AND REFLECTIONS IN A SINGLE PIPELINE ( 1. the .2. of a fluid is defined K=~ as a ~H= .7.which ca~ses a pressure rise of ~H = +(a/g) Vo. 1. (1.04 stopped. 1.15 that a=KOn the basis of Eq.l4) As the flow is completely 76 V = 1291 mls 2.16) 1291 9. 1. 1.10 Applied Hydraulic Transients Introduction 11 Rate of mass outflow = (Po + ~p) A (VO + ~V + a) (1. K.5 GPa.I8) + ~V) «a. Assume that the pipe is rigid.5)2 = 0. the pipe expands (in Fig. then the initial steadystate pressure head along the length of the pipeline IS H o. in which flow conditions ~re steady and at time t = 0. (1.~V ( 1..04 m/s ~V= .H is positive.5 m rise. I (1.S X 109 900 (Eq. In the next chapter. Behind this ~ave. P = 900 kg/rn ": and the bulk modulus of elasticity of the oil. K = 1. 1.18 is the velocity of waterhammer waves in a compressible fluid confined in a rigid pipe.12) which upon simplification becomes = 4" (0. the density of the oil. Because of this pres~ure nse. the initial steadystate pipe diameter IS sh. 1.16 becomes a=which may be written as =( 1.2.Let the distance x and the velocity V be positive in the downstream direction.11 ) The increase in the mass of control volume due to density change is small and may be neglected.04 m/s.(Vo + 6. .4/0. Hence.V=a Po The bulk modulus of elasticity. Eq.15) g ~p/Po It follows from Eqs. it is a pressure K 1... and a positive pressure wave propagates toward the reservoir. the rate of mass inflow is equal to the rate of mass outflow. Eq.13) a=v1 _ /1.18) Po Note that the expression of Eq.13 may be written as = 6. The sequence of events parts (Fig. . ~V ~p' Since the sign of 6.l Compute the velocity of pressure waves in a O.2aandb) As soon as the valve is closed. 1.196 = 2. 1.. V =0 = . we will discuss how this expression is modified if the pipe walls are elastic.14 and 1.p 6.2. the fluid is compressed thus increasing the fluid density. the valve is instantaneously closed.
1 t = 2L + . (e) L Conditions at .  = L Conditions at t ~I V=Vo V=O  I. ( f) L Conditions at t .' .RlIsllr"oir VOI"II~ RlIsllr"oir V=O I .. t +E V= Vo ~ l Conditions at t. Propagation of pressure waves caused by instantaneous closure of valve..2. ~_~ • I. Figure 1.:~~=1..   V=O L .2. a ~HY UIII roulC 9ro tJ..t V=O ______ 'F I. ( b) =~ a I' ~' Hy d roulC 9ro dll Iifill r RlIsllr"oir VOI""~ =.. (a) I I.I (HydrouliC grodll lillll _lr"..1 = 3L a  (c) Conditions at = .12 Applied Hydraulic Transients Hydroulic 9rodll linll r=" RlIsllr"oir L Introduction 9rodll lillll 13 ~ VoIIIII RlIsllr"oir VVo V·o l . (Continued) .AH" V=Vo I.!:... + E a Figure 1.II /.RlIsllr"oir r . (d) Conditions at 1!:.
friction losses neglected. the preceding sequence of events starts again at t = 4L/a.fro~ the pipeline into the reservoir with velocity .!« < i e. 1. At t = 2L/a. 2. and a negative wave propagates in the upstream direction. Thus.H to o 4L a 12L rim6 a Figure 1. Behind this wave. this process continues and the conditions are repeated at an interval of 4L/a. the fluid starts to flow towards the valve with velocity Va. the velocity IS changed from 0 to .2e and f) Since the valve is completely closed.Va. 's l. an unbalanced condition is created again at the upstream end. while Fig. Because of this pressure differential. 1.. a s~ction o~ t~e reservoir side is Ha while the pressure on an adjacent ~ectiOn In the pipe IS Ha + t:. the conditions in the pipeline are the same as during the initial steadystate conditions.'2L/a(Fig.t:. however. This interval after which conditions are repeated is termed the theoretical period of the pipeline. 3. Thus. 2L/a < i « 3L/a (Fig. Pressure variation at valve.e.. the pressure head in the entire pipeline is Ha. In other words. the pressure is reduced to H a . Because of this. 4. 'I 14 Applied Hydraulic Transients ~HYdraunC gradtJ hn6 Introduction 15 s~~rl  \.H.t:.l. and the fluid velocity is zero. the pressure head in the entire pipeline isHo .. Therefore. the Fluid starts to flow . pressure waves are dissipated due to friction (h) Conditions at t. 4L/a (Fig. and L is the length of the pipeline. and the pressure head is restored to Ha. and the fluid velocity is .. the flow velocity is zero.2 illustrates the sequence of events along the pipeline. Therefore..Va. As the valve is completely closed. along the entire length of the pipeline.H. and the fluid velocity is zero. At time t = 4L/a. the pressure is Ho ..H. As we assumed the system is frictionless. the conditions are unstable at the reservoir end when the wave reaches there because the pressure on.Vo.t:. a negative velocity cannot be maintained at the valve. (Continued) Reserfloir 1""". Now the pressure is higher on the reservoir side than in the pipeline.__ . the pressure head in the entire pipeline is Ho' and the flow velocity is Vo. on the upstream side) is Ha and the fluid velocity is . ~L Figure 1. At time t = 3£/a. 1..2. then at time t = L/a.3 shows the pressure variation at the valve end with time.3.Va' which causes the pressure to drop from Ha + t:..2g and h) As soon as this negative wave reaches the reservoir. a negative wave travels toward the valve such that the pressure behind the wave (i.2candd) .H. Figure 1... the velocity is instantaneously changed from .H. and the pressure head is H a + t:.Va to O.1:Fti R6stJrfloir r ( g)  Conditions at /" Hydraulic grade I~ RtJstJrfloir  Ha. L/a<t'. As the reservoir level is constant. In real physical systems. . the pipe is expanded.
Examples of such flows are flow in sewers following a rainstorm. w = frequency.16 Applied Hydraulic Transients Introduction 17 RIISllrvoir IIIvIII o !. Pressure variation at valve. Examples in which such transients occur are watersupply pipes. the system must be analyzed as a distributed system. and the fluid becomes stationary after a short time. are called combined freesurfacepressurized flows.. New York. any change in the flow conditions is assumed to take place instantaneously throughout the fluid. Transients in open channels may be divided into two types depending upon the rate at which they occur: (1) gradually varied flow. Such flows. and gastransmission lines.5. transients in closed conduits 2. transients in open channels 3. If the friction losses are taken into consideration. In the former case. and (2) rapidly varied flow. combined freesurfacepressurized transient flow. If wL/a is much less than I.5 CLASSIFICATION OF HYDRAULIC TRANSIENTS Depending upon the conduit in which the transient conditions are occurring.. and a = wave velocity.4. due to excessive transient pressures caused by operating errors and malfunctioning of equipment (After Bonin80 Courtesy of Ebasco Services Inc. then the pressure variation at the valve with time will be as shown in Fig. Mathematically. transients may be classified into three categories: 1. The analysis of transients in closed conduits may be further subdivided into two types: distributed systems and lumped systems. otherwise. . Japan.h Q Figure 1. and flow in the tailrace tunnel of a hydroelectric power plant following rapid acceptance of load on turbines. i. 1. the fluid is considered compressible. In the analysis of lumped systems.e. friction losses considered. power plant conduits. View of burst penstock of Oigawa Power Station.4. losses as the waves propagate in the pipeline. the fluid is considered as a solid body. Example of such a system is the slow oscillations of water level in a surge tank following a load change on the turbine.) the system may be analyzed as a lumped system 77. then Figure 1. L == length of the pipeline. such as flood waves in rivers. such as surges in power canals. Sometimes a freeflow becomes pressurized due to priming of the conduits during the transientstate conditions. it is referred to as a bore. the transients in the distributed systems are represented by partial differential equations. In the preceding expression. 1. whereas the transients in the lumped systems are described by ordinary differential equations. and the transient phenomenon occurs in the form of traveling waves. If the wave front in the rapidly varied flow is steep.
the intermediatestage flow. Opening. is termed transientstate flow. In other words. and the system is analyzed for transients caused by various possible operating conditions.6. accepting or rejecting load Vibrations of the vanes of a runner or an impeller.) 1.5. If the system response is not acceptable. caused by vacuum upstream of the burst section. Failure or collapse of a dam 7. the system layout and parameters are first selected. 2.18 Applied Hydraulic Transients Introduction 19 Figure 1.. View of collapsed section of penstock of Oigawa Power Station.7 SYSTEM DESIGN AND OPERATION To design a system. then either the system layout or . planned or accidental. closing. when the conditions are changed from one steady state to another. or of the blades of a fan 5. or "chattering" of valves in a pipeline Starting or stopping the pumps in a pumping system Startingup a hydraulic turbine. Sudden changes in the inflow or outflow of a canal by opening or closing the control gate 6. the transient conditions are initiated whenever the steadystate conditions are disturbed. in the settings of the control equipment of a manmade system and by changes in the inflow or outflow of a natural system. (After Bonin80 Courtesy of Ebasco Services Inc. Such a disturbance may be caused by changes.6 CAUSES OF TRANSIENTS As defined previously. 3. such as the maximum and minimum pressures are not within the prescribed limits. Common examples of the causes of transients in engineering systems are: 1. Japan. New York. (Continued) Figure 1. Sudden increases in the inflow to a river or a sewer due to flash storm runoff. 4. 1.
or it may be economical either to modify the operating conditions.20 Applied Hydraulic Transients Introduction 21 Figure 1. For example. To avoid catastrophes. if there are four parallel pumpingsets on a pipeline. similarly.5 and 1.6. 1. the system should be tested for various possible operating conditions. the tests for power failure should begin with one pumpingset and progressively increase to all four. For a particular system.6. the most commonly used terms were first defined.6 show the burst and the collapsed sections of the penstock of the Oigawa Power Station''? caused by operating errors and malfunctioning equipment. The system must be designed for various normal operating conditions expected to occur during its life. then the system should be analyzed for the expected range of various variables. if possible. This procedure is repeated until a desired response is obtained. Failure to do so has caused spectacular accidents 7883 and has resulted in extensive property damage and many times loss of life. However. A brief historical background of the development of the knowledge of hydraulic . the final aim is always to have an overall economical system that yields acceptable response. or to change the acceptable response. it is usually advisable to conduct the tests in a progressive manner. or various control devices are provided and the system is analyzed again.8 SUMMARY In this chapter. tContinued i the parameters are changed. If the data for a system are not precisely known. (Continued) Figure 1. Figures 1. During the commissioning of a newly built system or after major modifications. And. it is mandatory that the system be operated strictly according to the operating guidelines. a number of control devices may be suitable.
. 1789. M. XXXIII. Stuttgart. "Hydraulic Investigations. Sortis des Ecoles de Gand. E. 17. Assume the liquid is water." Trans. 9. "Concerning the Propagation Velocity of Waterhammer Waves in Elastic Pipes. 18931894. "De la Propagation du Son. S.J. Department of Civil Engineering. ANSWERS 1. Gromeka. Braun. Menabrea.. Bertrand's ed. Amer. R. pp. 16.. L. Paris. Sept. pp. Civ." Trans. 1878. New York.. 1875. Newton. Die Turbine. Wiss. History of Hydraulics. 20. Simin.5 A valve is suddenly opened at the downstream line such that the flow velocity is increased from 2 to 4 m/s. Hyd.2 Derive Eq.4 482. Weber. S.1877.22 Applied Hydraulic Transients 2.. Frizell. Introduction 23 transients was presented. France. I. Marey. N. (Hint: Apply Newton's second law of motion to the fluid volume. Braun.6577.. Soc. B.. April 1970. 1910.3? end of a lmdiarneter pipe1. II. vol. Water Works Assoc.. Switzerland. 118. 1909. Dover Publications. Young.R. Civil Engrs. 24.. no..:) :~ 24. 5664. that the pipe is inclined in a 2mdiameter to the . Monge. 1687. p. I. Riemann.. Soc." Phil. T. 14.. Braunschweig. p.." Report No. "Description of Some Experiments Made on the Providence. "Pressures Resulting from Changes of Velocity of Water in Pipes. vol. Problem 1. 8. 47. 1775.S. 1. 192.. 1876. Michaud. "Coups de belier dans les conduites. PROBLEMS 1. Kazan. Amer. 1898. "Teo ria generale del mota perturbate dell'acqu anei tubi in pressione. 21. 1.g dt in which L = length of the pipe and d V [dt = the rate of change of velocity with respect to time... Korteweg. 1904). 341424). i 25. 1869. vol. R. Petersburg.horizontal pipe conveying at an 10. Soc.3 1488 mls 1. vol. Amer. 1903 (French translation by Allievi. Joukowski.. 3 Assume the pipe is rigid.. Euler. 22. E. 19. 1858." Opera Postume Tomus Alter. 1. I. A. of Univ. Royal Society.. 1. 9. Compute the pressure drop due to the opening of the valve. 1900 (in Russian.1 Derive Eq.. 2939.. 4e annee. Weston. Euler..164186." Scientific Soc. P. d." Trans. Druckschwankungcn in Rohrleitungen . P. Carpenter..) 1. 5.S. translated by O. "Graphical Integration. Wiedemann. 353357. 7. 2. 6. Rouse." Bulletin de la Societe Vaudoise des lngenieurs et des Architectes. 102. of Civil Engrs.." Ann. Ing. pp. W. 65. "Menabrea's Note on Waterhammer: 1858.9 from first principles. of Kazan.!i . of St.4 What would be the pressure was instantaneously stopped rise if an initial steady discharge of 10 m /s at the downstream end of the pipeline of 12. Berlin. Book 2. des Ing. Revue de Mecanique. Engrs. F. "Note sur les effects de choc de I'eau dans les conduites. Lausanne. 13. London. and Ince. 1808. 1. ! ~ . 1904. of Mech. 3 and 4. The Principia. June 1898.1885.2 following system is frictionless. Arch. New Series. 1. 3rd Series. 3. Germany.. Imperial Academy Soc. E. Allievi. 1866. G. with time at the midlength of the pipeline instantaneous closure of the valve.. The chapter was concluded by a discussion of the classification and causes of hydraulic transients. Partielle Differentialgleichungen. Assume the and the pipe walls are assumed 14. MathematischPhysische Klasse. 1963. 15. 23. Celestial Mechanics.. "Principia pro Motu Sanguinis per Arterias Determinando. C. ed.. Paris. pp. italiana. 1..S 301.6 Prove that if the fluid is incompressible rigid. TheoryofSound. 4 volumes." Annalen der Physik und Chemie. Soc. Queen's University at Kingston. May 1883. 5. I." Ann. 525542. D." Berichte iiber die Veriiandlungen der Koniglichen Sachsischen Gesselschaft der wissenschaften zu Leipzig. and expressions were derived for the pressure rise or drop due to an instantaneous increase or decrease in the flow velocity.81 m '. Laplace. France. Rayleigh. then the pressure rise L dV tlH= . 221224. Bowditch translation.." Mhnoires de l'Acad. Marey. pp. 'c • . B. 5.. vol. pp. L. L.7 Plot the pressure variation shown in Fig. Div. 26. pp. Resal." Comptes Rendus Hebdomadaires des Seances de L 'Academic des Sciences. "Theorie der durch Wasser oder andere incompressible Fliissigkeiten in elastischen Rohren fortgepflanzten Wellen. F. J. Water Works to Ascertain the Force of Water Ram in Pipes." Travaux du Laboratoire de M.Mecanique Analytique.. 1759. pp... Engrs. J.1. Organ der turbinentechnischen Gesellschaft . London." Jour. "Mouvement des Ondes Liquides pour Servir a la Theorie du Pouls. Wittwer. 12.. 18. Ontario. 1878. Soc. 1976. Royal Soc. 39. Wood.86 m REFERENCES 1. Proposition 4446. L.. "Note sur les petits mouvements d'un fluide incompressible dans un tuyau elastique.. vol. pp. 4. Proc." Journal de Mathematiques Pures et Appliquees. "History of Waterhammer. Amer. Leipzig. S. I. E. 342344.8 assuming angle (J. 238. 1. Berlin. Amer.. and Dec. nos. H. W. JulyDec. U.. Trans.. Mem.3 Compute the wave velocity seawater. Anderson. no. 1788. Paris. vol. "Experiments on Waterhammer. Jan. M. Lagrange. Canada. vol.. H. "Ueber die Fortpflanzungsgeschwindingkeit des Schalles in elastischen Rohren. Etude des moyens employes pour en atteneur les effets. L. 5.
Halmos. Rich.. "Valve Stroking to Control Waterhammer. Civil Engrs. Paynte. vol. G. V.1963. J. . Illinois.. Mech.ts. "Bestimmung von Schliesstuncuonen welche den kleinsten Wert des maximalen Druckstosses ergeben. Germany. "Simple Graphical Solutions for Pressure Rise in Pipe and Pump Discharge Lines. C.. France. Prasil. Engrs. vol.. Jan. Wood. Bauztg. "A Study of Waterhammer Including Effect of Hydraulic Losses. 1910.. Soc. R. "Present Trends in Surge Tank Design. L. 65. 6. "Pozzi Piezometrici e Stabilita di Regolazione.. Jaeger." Trans.. Inc.. Water Engrs. Lausanne. F. "Druckstosse in Pumpensteigleitungen. Traite des Turbomachines. Dec. C.1967. E. Cill. 53b. 63. 53a. Civil Engrs. of Mech. Rich. P. "The Analysis of the Dissipation of Energy in Waterhammer " Proc. 28. Mech. Inst.. c. Engineering Institute of Canada. Hydraulic Transients." Paris." Schweiz.. 40_ "Symposium on Waterhammer. Gibson. Amer. Bergeron. Mech. C. New York. R. "Speed Changes of Hydraulic Turbines for Sudden Changes of Load. Hydraulic Transients. 31. "The Differential Surge Tank. 59. 1943.no. Theorie des Chambres d'equilibre." Trans. A. E. 41." Trans.. pp. "Waterhammer Analysis by the LaplaceMellin Transformations. 60. L. "Ueb. of Canada Conference. Amer. Vogt. Soc." Comptes Rendus des Travaux de la Soc. 19441945.. M. Lui. Amer.. 1956. of Mech. Engrs. 59. de France. Amer. Paris. Stuttgart.. 1915. R. N. 73. Inst. vol. Parmakian. A. Aug. M. 33. in partial fulfillment of the requirements for the degree of Doctor of Philosophy. 340401. of Civil Engrs. V. 1929). 45. S. E. 45.. Denver Colorado April 1966. Leaute. 1953. 62.. "Transient Analysis of Certain Nonlinear Systems in Hydroelectric Plan. Wood. 1937. 65. 58. Soc. des Ing.. 64. 119. McGrawHill Book Co. L. Paris. C. 38. Soc.. pp. 1880. Soc. T. Druckschwankungen ill Druckrohrleitungen. Enke. 1956. L. 52." J. Engrs. Soc. Munich. Gibson. 1908. 39... June 1933. March 1963.. Thoma. Lupton. Gray." Soc. 1939. Streeter. 48. Springer.. "Variations de regime dans les conduites d'eau. Paris. Jaeger. "Bermerkungen zur Theorie der Druckschwankungen in Rohrleirungen. 42. and WYlie. M. Dunod. and by Paynter. Munich. 181]83. M. Ann... L. vol. vol. 1st Ed. "The Application of Heavisides Operational Calculus to the Solution of Problems in Waterhammer.pp. 56. Bergeron." Jour. 30. Amer. Waterhammer Analysis. Inc. F." thesis presented to the University of Michigan. "Waterhammer Pressure in Compound and Branched Pipes. in partial fulfillment of the requirements for the degree of Doctor of Engineering. vol. Schnyder. "Recherches theoriques et experimentales sur les oscillations de l'eau dans les chambres d'equilibre. 66." Dunod." Proc. 47. Engrs. l.. Eydoux. ] 951 (Dover Reprint). H. Soc. Amer.." Trans.. paper 274. Cabelka... 1963). "Water Column Separation in Pump Discharge Lines. R. W. pp.. 1926. 1959. R.. B.. Engrs. J. 7281. Engrs." Trans. L. 1900. "Water Column Separation. Mech. ' . 1943. Schnyder. vol. 50. 1923. and Franc. in partial fulfillment of the requirements of degree of doctor of philosophy. Soc. 128. Mech.. vol.1953. V. 48. Angus. Engrs. C. "Closure Characteristics of a Valve with Respect to Waterhammer. W. pp_ 14911524 (see also discussions by Gray. Camiche1. of Karlsruhe Germany. vol. M.. Rateau." Trans. NoveDec. pp. and A mer. Soc.. 89. J. pp. "Air Chambers for Discharge Pipes. Massachusetts. Due.. "Etude Theorique et Experirnentale des Coups de Belier . Streeter. SWitzerland. Schwingungen ill den Zuleitungsund Ableitungskaniilen 1'011 Wasserkraftanlagen. Soc. Wasserwirtschafl.." Trans. for Hydraulic Research Montreal. 1913 (English translation by E. 1957." A tti Collegia lng. 54. and Gariel. Engrs. vol. Constantinescu. Allievi.. 1959.. Engrs. and Kerr. Paris... 11761194_ ' 57. Engrs.... Soc. 15301533. 168. and Lai. vol. 0. 83. Ami'r. 6470. 1956. A mer. 61. Engrs. Allievi... 29." Proc. pp. Ann Arbor. Arch.. Englewood Cliffs N J 1955 (Dover Reprint. vol. J. vol. F. Amer_ Soc. ' ." Energia Elettrica. 43. Soc. R. H.. 407497.. 1938.. pp. Amer. England. E. Clame. "Pressures in Penstocks Caused by Gradual Closing of Turbine Gates. translated from German by Wolf. Richard. N. pp. ' 55. 209262. D." Trans. D.760805.. 46. D.. thesis presented to Massachusetts Institute of Technology. "Oscillations in Closed Surge Tanks. Blackie & Son Ltd.. 1929. Civ. 5 and 6. R. C. 1951. Symposium on Waterhammer. Soc. vol. New York.. 651659. Review.1932. pp." wasserkraft u.. "The Gibson Method and Apparatus for Measuring the Flow of Water in Closed Conduits. Engrs. PrenticeHall. of Mech. nos. ' 67.. no. 1935. 94. Streeter. Zur Theorie des Wasserschlosses bei selbsttdtig geregelten Turbinenanlagen Oldenbourg .19191920.. 90.." Sulzer Tech. Amer. "Memoires sur les oscillations tongues periodes dans les machines actionnees par des rnoteurs hydrauliques. no. Amer." presented at Amer. Johnson. 1938. 27. pp. Soc.. "Teoria del colpo d'ariete. vol. 59. vol. G. vol. F. 1953. Civils de France. pp. F. Soc. 283286. 707775. G.. Eighth Congress. Hvdrotech. Engineering Fluid Mechanics. vol. Amer. 72. 52. 1919. Frank. 36. discussion of Ref 34. Braun. and Gaden. "Waterhammer Analysis Including Fluid Frictlon " Trans.. Nov.r Druckstosse in Rohrleitungen. 1937.." Jour.. 27... Canada. 1928.24 Applied Hydraulic Transients Introduction 2S 27. G. 91103." Schweizerische Bauzeitung. E.O. Engrs.pp.. Escande. B. 1919. 15331538). Heft 5.. A. 1954. 29. Angus. "Graphical Analysis of Pressure Surges in Pumping Systems. Strowger. Allier. 3. Springer. 6A1 to 6A23. Trails.. 1962.. Ruus. 3966. 30. Mech. 70. Jan. 16.. Engrs. vol. E. 34.. Dec. L. M. Div. 0. 78. 68.. Chambres d'Equilibre. 1937. Rouge et Cie. London. Evangelisti. 1920_ 32.. Annual Meeting. Soc." Wasserkraft lind wasserwirtschaft." Proc. 48. 28. 41." Trans. "Optimum Rate of Closure of Hydraulic Turbine Gates. Hvd. Amer. pp.. Berechnung und Konstruktion des Wasserschlosses. Soc. A. 49.707713. Ruus. 343392.1950. Engrs.. Proc. Engrs. 1926. M.. 44.~' H". Binnie." Jour. 37. pp.. L. vol. Nov. 1931. 271273. pp.." Amer. Michigan. Civ." thesis presented to the Technical Univ. 71. pp. of Mech. Berlin. Lcwy. 1923. J. M. Cahier XLVII.. 1926. 35.. D. Chicago. Gardel. and Schuller.1951. R. "The Theory of Waterhammer. pp. A. Ecole Polytechn... Mech. 51. R. Civ." Trans.. Engrs. Amer. des Mines de Roumanie. Mech.. "Wasserschlossprobleme. a . Mech. Feb.. 4954.Engineering Inst. Soc. 1937. International Assoc.. L. 69. McGrawHili Book Co. "Method graphique generale de caicul des propagations d'ondes planes.
Hydrauliques. The conduit walls and the fluid are linearly elastic. of Mech. 76. For computing frequencydependent friction. of Water Hammer Waves.. Inst. Works Jour.2 DYNAMIC EQUATION We will use the following notation: distance. Let us consider a horizontal element of fluid having crosssectional area A and length OX. The validity of this assumption has not as yet been verified. 1963. McGrawHili Book Co. London. 1937. 1948. Paynter. 'The dents Occurring 82. 1976. ~~!~dhry. are considered positive in the downstream direction (see Fig. concrete and wooden pipes. and H is the piezometric head at the centerline of the conduit above the specified datum. Boni'n. Amer. In the xdirection. 631640. Discussion 011 of "The Velocity Symposium Surges in Pipelines. Jaeger.. "Surge and WaterHammer Problems.. C. H. In this chapter. W. Flow in the conduit is onedimensional. 2. If the piezometric head and the velocity at distance x are H and V. pp. 75. L. Soc. • 2. Soc. of lnci vol. EQUATIONS OF UNSTEADY FLOW THROUGH CLOSED CONDUITS 77. A.." This is true for most conduits such as metal.." by Pearsall. Dec. Power.. "Literature Survey of Water Hammer Incidents in Operatin~ Nuc~ear Power Plants. voL 81. V. Pittsburgh. T. University quirements for the degree of doctor of philosophy. However. A mer. Trans.. pp... Basic Engineering.. London." Report No. 2. dans les Installations . Mech.. Feb.. these procedures are too complex and cumbersome for general use. Water Level Displacements M. c. Jour. 2. discharge. Kerensky. Zielke! has developed a procedure for laminar flows. nos. 78. New York. • 2. 25. 81. Westinghouse Electric Corporation. WCAP8799. and Hirose'' has proposed an empirical procedure for turbulent flows. Fluid Mechanics. Engrs.. then their corresponding values at x + ox are H + (aH/ax) ox and V + (a v/ax) ox.. Nov. Soc. pp. 1.. 1959.. 19651966. Piping Systems. Trans. Vancouver. 79.May."? and the velocity distribution is uniform over the cross section of the conduit." thesis presented to the Canada.. Pulling. and lined or unlined rock tunnels. "Les Phenomenes d'AutoOscillation Hermann. in partial fulfillment of the re1970. 1966. . p. 1953. 80. April 1960. pp. Q. Pt 3 E.." Electrical Analogies and Electronic Computer Symposium. Amer.. 23. Unsteady flow through closed conduits is described by the dynamic and continuityequations." Rev. Y." Civil Engineering Pub. C. Streeter. 3. and flow velocity. V. Pennsylvania. Engrs. ofMech. the derivation of these equations is presented. A mer.1.. 180. S. vol. Civil Engrs.. C. 4th Ed. Soc. and methods available for their solution are discussed. G. Jour. Marris. 118. Rocard. vol. Engrs. "Water Hammer Effects in Power Conduits. Jaeger. three 27 Engrs. 111119. M.1 ASSUMPTIONS The following assumptions are made in the derivation of the equations: 1.1). i.i'stress is proportional to strain. W. x. 9621009. "WaterHammer Damage to Oigawa Power Station. and we will not discuss them further. H. p. Formulas for computing the steadystate friction losses in conduits are valid during the transient state. "Resonance in Pressurized of British Columbia. Mech. 85. Theory of Resonance in Pressure Systems.e.26 Applied Hydraulic Transients CHAPTER 2 74. 83. Engineering [or Discussion . respectively. 500503. within a conduit as shown in Fig. 24. "Large " in the Simple Surge Tank. 446454." in Hydropower Systems. Basic Engineering. Paris. vol.
2.oro xl" + x IX) .1 through 2.7 into Eq. If "f = specific weight of the fluid.4) Substitution of the expressions into Eq. F2.1) Eq. FJ. 2. Notation for dynamic equation. 2. then FJ = "fA (H . 2.5 and 2. f= friction factor. F2. A = crosssectional area of conduit.10) .9b into Eq.5) second law of motion. The resultant force.4 yields F= "fA According to Newton's for F1. and S. . 2. F. and S from 2 Eqs. 2. FJ and F2 are forces due to pressure while S is the shear force due to friction. of the element dV dt (2.7) of Eqs.9a) dV av =+ dt at Substituting av V ax fV2 (2.3 ~ x Q I 1 x + 8x Q aH "Y ox . QI I +1 z x I"" I ~ 8x ±::9I I I rDatum .2) + V+g+=O at av av ax aH ax 2D (2.9b) (2. and D = diameter of the conduit. forces.6) H H+ (a) For the fluid element under consideration. Force + ~~ ~~ 8x 8x = Mass X Acceleration. (2.8 and rearranging.[ i in which g = acceleration due to gravity..fV8 rrD ox ax g (2. Mass of the element = ~ A ox} = s ( b) Free body diaoram Acceleration Substitution .6 and division by "fA Sx yield dV dt We know from elementary Figure 2. are acting on the element. F2 = "f ~ z + aa~ ox)A (2.3) 1 II 1 Positive I x I V.. and z = height of conduit above datum. ?"I Equations of Unsteady I Flow Through Closed Conduits 29 losses.28 Applied Hydraulic Transients Instantaneous hydraulic orode line 1 f.z) or = g  aH fV2  ax 2D (2. If the DarcyWeisbach then the shear force formula is used "f fV2 for compu ting the friction " "I 1 S=! g rrDox 8 (2.8) calculus that the total derivative dV dt =+at av aVdx ax dt (2. acting on the element is given by the equation (2.1.
14) E=OE oa of expressions (2.16) r has increased to r + or. To account for the reverse flow. 2. and a r.18) for in which E = Young's modulus of elasticity.13) [f the conduit walls are assumed linearly elastic. So.ote at e Since the radius r ap r (2. If correct values of m and b are used. and b depending upon the formula employed. we are neglecting the elongation or shortening of the fluid element due to Poisson ratio effects. then the last term of Eq..17) t"out = in which r = radius of the conduit. 2.8.= . 2.21)  . I Inflow . Let us first consider the volume change. 2. Ot"in.. OE= the change in strain = V 1Tr2ot (2." the term V(a Vlax) is significantly smaller than the term vial. By writing Eq. 23 CONTINUITY EQUATION of fluid I.3. The pressure change. m = I and b = 3. ing time (V + ~: ox ) 1Tr2 0 t (2.r rearranging.11 ) I I I I I I I I I Outflow Eqs. caused by op may be written as pr (2.1 0 in terms of discharge. 2.a . 2.l2) or r (2. for the DarcyWeisbach formula.. Q.. m. This pressure change causes the conduit walls to expand or contract radially and causes the length of the fluid element to decrease or increase due to fluid compressibility (see Fig. Sp . Vin. the expression V2 in Eq. 2. or=ot ap r2 eE due to the radial expansion (2.~I""" .16 and 2.11 could be written as kQ IQ 1mIDb. Therefore. we obtain aQ f ar+ gA sn + 2DA ax In QIQI = 0 (2. . and 2.l0 may be written as VIVI. while.10.19) I " i i I . 2. For example..20) or contrac at ~e change in the volume of the element tion of the conduit is O¥y = 21Tr ox or (2. Notation I equation.e. The radial or hoop stress. i. t"out' during time interval Sr are Vin So = op . the former may be neglected. L6~_J Figure 2. in which IVI is the absolute value of V.18 yields oa "_. for the HazenWilliams formula.30 Applied Hydraulic Transients Equations of Unsteady I Flow Through Closed Conduits 31 In most of the transient problems. dur ot is (2.. E = (aplat) ot (rle) 8rlr or (2." the results are independent of the formula employed.17 into Eq.87. If a general exponential formula had been used for these losses. a.15) Hence.85 and b = 2. then The increase in the fluid volume. 2. 7.2). the DarcyWeisbach and the HazenWiIliams formulas would give comparable results.. for continuity the pressure p is given by the equation" a=e in which e = the conduit wall thickness. 2. Substitution and OE from Eqs.2.11. the change in hoop stress. Let us consider the control volume shown in Fig. 2. in a conduit due to *To simplify the derivation. The volume inflow. the DarcyWeisbach formula has been used for calculating the friction losses. and outflow. m = 1. Any reader interested in the derivation of the continuity equation including these effects should see Ref. ot"n due to the radial expansion or contraction" of the conduit. with the values of k.5. during time interval ot is (aplat) St. as derived above.2.
a. 2. However. Eq.11 and 2. are a set of firstorder partial differential equations. and two dependent variables. are characteristics of the conduit system and are timeinvariant but may be functions of x. and the continuity equation. for the change The initial volume of the fluid element ¥ = rrr2 8x The bulk modulus of elasticity of a fluid. 2.11 and 2. any reader not interested in the mathematical details may proceed to Section 2. These equations may be further classified as elliptic. 2.23 and noting 8¥ c = ap at 8t rrr2 Sx K (2.4 GENERAL REMARKS ON DYNAMIC CONTINUITY EQUATIONS AND Let us now derive an expression pressibility of the fluid. a2 aQ aH at (2. A and D. The friction factor fvaries with the Reynolds number. 2.30 involve only the first power of the derivatives.22) 8¥c.14. 2.30) of waterhammer The eigenvalues. '" of matrix B determine The characteristic equation"? of matrix B is It will be shown in the next chapter that a is the velocity .6.32 Applied Hydraulic Transients Equations of Unsteady Flow Through Closed Conduits conditions 33 Substituting for Sr from Eq.32) terms. the equations are called quasilinear.29) p = mass density of the fluid.24) that 8p = (ap/at) bt. 8¥" and 8J7<c from Eqs.25 into the above equation and division by rrr2 8x 8t yield =ax K at eE or av 1 ap ap at (2. f is considered constant herein because the effects of such a variation on the transientstate conditions are negligible.28) (2.)=o ax at eE K Let us define (2. Eq.11. Eq. due to com2. 2. and substituting Q = VA. laboratory tests have shown that it is significantly reduced? by reduction of pressure even when it remains above the vapor pressure. there are two independent variables. depends upon the characteristics of the system.11 and 2.33) . 2. 2. 2.30 may be written in matrix form as 2.25) then it follows from the If we assume that the fluid density law of conservation of mass that remains constant. Other variables. Eq. In these equations.22. Expressions for a for various conduits sented in Section 2. x and t.27) in which av + ap(~+.26) Substitution of expressions for 8¥in.30 now follows.30. K. Although the wave velocity. and 2r The dynamic equation. is de fined I (2.28 becomes +=0 gA ax o the type of the set of equations.!. and support are pre 8x in volume. Discussion about the type of Eqs.20 yields 8¥ r = 2rr  ap r3 +Bt at eE waves. parabolic.23) as 8p K By substituting becomes = 8¥cW (2. or hyperbolic as follows: Equations 2. G= rearranging the fQIQI} { 2DA (2. (2. Since the nonlinear terms in Eqs.31 ) a2 in which K =p[1 + (KD/eE)] Noting that p and (2.2 _ a 2 =0 (2. 2.24 for ¥ from Eq. Q and H.. = pgH.5. (2.
in addition to the bulk modulus of elasticity. The fluid compressibility is increased by the presence of free gases. E = Young's modulus of elasticity of the conduit walls. various graphical7. (2. The bulk modulus of elasticity of a fluid depends upon its temperature. Therefore. Elastic properties include the conduit size. Expressions for IjJ for various conditions are as follows: 1. and hence Eqs. Because each boundary condition and each conduit section are analyzed separately during a time step. such as the implicit finitedifference method 18 and the method of characteristics. To overcome this. Therefore. Halliwell/" presented the following general expression for the wave velocity: a =1 p[1 + (~/E) 1/IJ (2. which must be solved by an iterative technique. ThickWalled Elastic Conduits a. and the resulting algebraic equations for the whole system are then solved simultaneously.5. the time step cannot be increased arbitrarily because it results in smoothing the pressure peaks. and it has been found2s that 1 part of air in 10. A=±a (2. partial differential equations.1824 In the implicit finitedifference method. A cJose. in which 1/1 is a nondimensional parameter that depends upon the elastic properties of the conduit. *For a derivation of expressions for the wave velocity in gasliquid mixtures.1 1. Conduit anchored against longitudinal movement throughout its length 1/1 = 2(1 + v) R2 + R~ 0 2vR/f in detail in Chapter 3. Pearsall " has shown that the wave velocity changes by about I percent per 5°C. larger time steps can be used.5 METHODS FOR SOLVING CONTINUITY AND DYNAMIC EQUATIONS As demonstrated previously.1 and 2.17 they are not presented herein because their programming is difficult. These methods are approximate and cannot be used to analyze large systems or systems having complex boundary conditions. 8. wall thickness. Rigid Conduits 1/1=0 (2. . Laboratory? and prototype tests2S have shown that the dissolved gases tend to come out of solution when the pressure is reduced. and K and p are the bulk modulus of elasticity and density of the fluid.34) Since a is real. However. the velocity of waterhammer waves depends upon the elastic properties of the conduit.34 Applied Hydraulic Transients Equations of Unsteady Flow Through Closed Conduits 35 Hence. respectively.33 form a set of hyperbolic partial differential equations. Further prototype tests are needed to quantify the reduction in the wave velocity due to reduction of pressures.2.35) ?7. This causes a significant reduction in the wave velocity. the dynamic and continuity equations are quasilinear. both eigenvalues are real and distinct. . 16.11 and 2. of the fluid. = the external and internal radii of the conduit. hyperbolic. pressure. the partial derivatives are replaced by finite differences. by neglecting or linearizing the nonlinear terms. R~ .dfor~~solution of ~hese equations is impossible. The moduli of elasticity of commonly used materials for conduit walls and the bulk moduli of elasticity and mass densities of various liquids are listed in Tables 2. the external constraints include the type of supports and the freedom of conduit movement in the longitudinal direction. 2. The method has the advantage that it is unconditionally stable. a combination of the implicit finite difference and the method of characteristics23 may be used.36) 2. which results in economizing computer time. * Solids in liquids have similar but less drastic influence. this method is particularly suitable for systems with complex boundary conditions. However. this involves a simultaneous solution of a large number of nonlinear equations.000 parts of water by volume reduces the wave velocity by about 50 percent. The disadvantage of this method is that small time steps must be used to satisfy the Courant conditionl8 for stability.R. 2. 2. The analysis by this method becomes even more complicated in systems having complex boundary conditions. unless they are compressible.12 and analytical 1315 methods have been developed. as well as on the external constraints.Rt I R~ . Depending upon the size of the system. the wave velocity for a positive wave may be higher than that of a negative wave. even when it remains above the vapor pressure. Although some of these methods have been programmed for analysis on a digital computer.2.37) in which v = the Poisson's ratio and Ro and R. This is discussed K. the partial differential equations are first converted into ordinary differential equations. which are then solved by an explicit finitedifference technique. However.6 VELOCITY OF WATERHAMMER WAVES An expression for the velocity of waterhammer waves in a rigid conduit was derived in Section 1. see Section 9. Details of this method are presented in Section In the method of characteristics. We will discuss techniques thatare more suitable for computer analysis. and the quantity of undissolved gases. and wall material.
Tunnels Through Solid Rock Halliwell'" has derived long expressions for 1/1 for lined and unlined rock tunnels.42.8 6. Usually.39) 4. and 33.8 5.2 1.33 0.R.570 900 999 1.43 1.05 1. • 'To convert E into Ib/in.32 4. Steellined tunnel in which D = conduit diameter and e = wall thickness. using Halliwell's expressions for practical applications is unwarranted. 12. in our opinion. sea 15 0 15 20 20 15 20 15 880 790 1.24 0. =12 e (2. Conduit anchored against longitudinal movement its length (2.21 0. E** (GPa) 6873 24 78110 80170 1430 107131 4673 4. multiply the values given in this column by 145.25. Therefore. K:I: (GPa) 1. det. fresh Water. Conduit anchored against longitudinal VI = movement at the upper end R2 .44 0.sRi + v(R~ .3Rt)] R2 _ R~ o I (2.R~ 0 I 2 [R~ + l.044.44) . the rock characteristics cannot be precisely estimated because of nonhomogeneous rock conditions and because of the presence of fissures. :l:To convert K into Ib/in. ThinWalled Elastic Conduits a. + + v) throughout (2.8 2.10. 1/1= DE GD+Ee (2.43) in which G = modulus of rigidity of the rock. Unlined tunnel 3.1.025 'Compiled fro'm Refs.ermine t~e specific weight of the liquid.32.754. multiply the values given in this column by 62.427 X 10'.4 Density.36 0.42) b.260 804 13. * Temperature (Oe) Poisson's Ratio 0.42.25 .25.46 0. Liquid Benzene Ethyl alcohol Glycerin Kerosine Mercury Oil Bulk modulus of elasticity and density of common liquids at atmospheric pressure.40) E=G I/I=l} (2.* Material Aluminum alloys Asbestos cement..518. pt (kg/m3) Bulk Modulus of Elasticity.36 Applied Hydraulic Transients Young's modulus of elasticity and Poisson's ratio for various pipe materials. tre b.33 0.0 2.75 6.32 26.038 X 10'.41) 1/1 = D e (1. Instead. 12.75 50 55 24.7 1. multiply the values given in this column by 145.25 0.27 Water.Compiled from Refs.19 2.15 0. Conduit anchored against longitudinal 0. of Unsteady Flow Through Closed Conduits 37 Table 2.v) c. a.0 0.33 0. the following expressions based on Parmakian's equations 7 may be used.'.34 0. and 31 .27 0. b.6 Equations Table 2. Conduit with frequent expansion joints VI . Conduit with frequent expansion joints VI = 2 (R~ R~ Ro .2.38) c.817 200212 1.'. . transite Brass Cast iron Concrete Copper Glass Lead Mild steel Plastics ABS Nylon Perspex Polyethylene Polystyrene PVC rigid Rocks Granite Limestone Quartzite Sandstone Schist Modulus of Elasticity. in Ibrlft'.28 movement at the upper end (2.5 2.038 X 10'.28 0.
to allow for any cracks in the concrete pipe. Thorley and Guyrner+? have included the influence of the shear force on the bending deflection of the thickwalled (lIe 20) rectangular conduits while deriving the equations for the wave velocity. 2.5(6 . and K and p for water were taken as 207 GPa. 2. Reinforced Concrete Pipes The reinforced concrete equivalent thickness 7 pipe is replaced by an equivalent steel pipe having cross section: or a t lIC walled conduit having a square k I/J=I IS (/)3 e I E +.47) and ' (2. b = width Pipe No.46) (3 = 0. and E. and d = depth of the conduit (shorter side).35.71 353 . (2. Usually the value of E. The values of E for steel G for concrete. = expression Base~ on .71 5.05 is suggested.19 GPa. respectively.the eq~ations presented by Thorley and Twyman. varies from 0. a value of 0. owned and operated by British Columbia Hydro and Power Authority are listed 10 the following table: ' 7." Having computed ee.50:) + 0.5 Diameter (m) 6. Noncircular Conduits The following expression for I/J is obtained from the equation for the wave velocity in the thinwalled rectangular conduits derived by Jenkner''" by using the steadystate bending theory and by allowing the corners of the conduit to rotate: Pipe No.48) 6. 20. However. in which e = wall thickness. The wave velocity is then computed from Eq.respectively. 2. respectively. (d/b)].5 (dlb)3[6 .5 (bld)2].e) = inside G = shear modulus of the wall material. 3 and 999 kg/m . = mean width of one of the flat sides of the hexagonal to.55 Wall Thickness. 27 show that Eq. 0: = [1 + (dlb)3]1[1 + of the conduit (longer side). = ec = thickness of wood staves. 2. the wave velocity may be determined from Eq. Polyvinyl Chloride (PVC) and Reinforced Plastic Pipe Investigations reported in Ref. provided a proper value of the modulus of elasticity for the wall mate rial is use d .1 < D == e 0. 2. 36.019 6. and As and Is are the crosssectional area and the spacing of the steel bands..45 using E.( 1 +.06 to 0. 2. 8. WoodStave Pipes The thickness of a uniform steel pipe equivalent to the woodstave pipe is determined 7 from Eq. I 2 Length (m) 244.As and Is are the crosssectional area and the spacing of steel bars. = ratio of the modulus of elasticity of concrete to that of steel. the waterhammer wave velocity in each section of the penstock was determined as follows.35 can be used for computing wave velocity in the polyvinyl chloride (PVC) and reinforced plastic pipes.38 Applied Hydraulic Transients of the steel liner and E =modulus of elasticity of following Equations expression of Unsteady Flow Through Closed Conduits 39 in which steel.) e 2G dimension of the conduit (2. (1.7 CASE STUDY The data for the steel penstock of the Kootenay Canal hydroelectric power ~Iant.1. (mm) Remarks Expansion coupling at one end Encased in concrete 19 22 (2.I? the followin IS obtained for I/J for a thinwalled hexagonal conduit: g I/J in which I = 0. From these equations.35 using the modulus elasticity of steel.45) in which ec thickness of the concrete pipe. the in which ~or conducting a transient analysis. e =thickness is obtained fl' 5.0385 (.Y section.7 GPa.
March 1968.(1. 1961. was td presen e .62) = 2.5 m.2. John Wiley & Sons. Strength of Materials. The 20mm reinforcing bars have a spacing of 0.022 = 9.2mdiameter copper pipe having a wall thickness of 25 mm is conveying kerosene oil at 20°C from a container to a valve. Third Edition. In this chapter. REFERENCES I. Expressions f~r the velocity of waterhammer waves in the closed conduits were presented.19 X 109 999(1 1410 2. H..2.2. Advanced Mechanics of Fluids.3. Inc. 992 mls 3. Handbook of Fluid Dynamics. 0.8 m/s. the derivation of the dynamic ~nd . 1941. SUMMARY V. is anchored at the upstream end 3. (ed.30) the velocity of waterhammer waves in a 3.24 GPa. .). /~o~~:tions and various methods available for their solution were dl~cussed. Streeter..15m wall thickness.55 X 207 X 109 = 20.. New York.. McG~awHill Book Co. li h b li partial differend t ated that these equations are quasi.2. New York.10 shows the power conduits of an underground hydroelectric power station. Basic Engi .. A 0. 5. at what velocity would the pressure waves propagate in the pipe? Assume the pipe is anchored at the upper end.2.4) m/s.2 Equations pipe No. L. Figure 5. is embedded in a concrete dam 2. 2. 1232 + 0.05mdiameter having a wall thickness of 25 mm if it: a= ~ == 2.7 X 109 X 5. Equations of Unsteady Flow Through Closed Conduits 41 As the pipe is anchored PROBLEMS e (Eq. Compute penstock equation considering the conduit is inclined at angle 8 steel l/J D == . 1.1. Inc. Pipe No.19 K p[l + (KIE)l/JJ (Eq. an . (ed. Determine the velocity of waterhammer waves in a reinforced concrete pipe having 1..continuity .). for a steellined tunnel may be used to compute the wave velocity in 2. has expansion joints throughout its length. Meek Engrs. E 207 = 0.5. S. Jour.2.3.55 + 207 X 109 X . McGrawHili Book Co.0106 X 335.. 2.25 ..v) == 353(1. If the valve is closed instantly.mear. Derive the dynamic to the horizontal. 2. and carrying water. Second Edition. W. Assume modulus of rigidity of rock = 5.19 X 109 999(1 a= == 694 + 0.4.0. Van Nostrand Pipe Flow. New York. Soc.4 . Part 2. V." Co.41) 2. 9B mls mls a== 2. 978 mls 2.0106 X 9.44) ANSWERS 2. 4. 1966. Fluid Mechanics. 1959.62 (Eq. and the pipe has expansion joints throughout its length.0106 2. yper OIC. Compute the wave velocity in each section of the conduit.25m diameter. New York. DE l/J==CD+Ee 5.40 Applied Hydraulic Transients at one end.. 1413 mls 2. Zielke.equati~nsItwere d the assumptions used in these derivations were dlscusse. L. A mer.35) 2. Timoshenko.25 == 335. Rouse.4. Streeter. 3. "FrequencyDependent Friction in Transient neering.. I.
"A Method of Programming Graphical Surges in Pipelines. . Soc. 92. E. Joukowsky. E. Hydraulics Div... Engrs vol 97 March 1975. England. Soc. E. Waterhammer in Hydraulics and Wave Surges in Electricity. 18. New York... Hydraulic Models. and Ippen. and Lai. Dover Publications. 26. 1925. Hydraulics Div. Dec. S. 88.. Tedrow. M. I and 6. R. Inc. "FrequencyDependent Wall Shear in Transient Fluid Flow Simulation of Unsteady Turbulent Flow. L." Jour. vol. Fluid Engineering. Mech. No. 1968. A. 2252. J. von Druckstossprobleme mit Hilfe von Kunststoffrohren.. 23. V. for Hydraulic Research London.. Hydraulic Transients. C. K. L. 865873. J.'' vol. ' 30. ''Waterhammer. Engrs. 1960. Soc. Inc. 1227. 95.. Ltd. V. Harding. of Bends on Fluid Transients Propagated ' of Mech.." . 1963. D. lnstitut Berlin 1962.. Trans. Soc. "Numerical Methods for Calculation of Transient Flow. Wood. pp." L 'Energia Elettrica. Mo~k. . New York. and Stoner. Soc. Nos. "A BoundaryLayer Theory for Transient Viscous Losses in Turbulen t How. ... D." Proc. Trans. 1972. 22. in partial fulfillment of the requirements for the degree of doctor of engineering. R. 17.. 121. Perkins. (ed.). E. Hydraulic Transients. of Hyd. Streeter. Inst.. "The Use of the Computerized Graphical Method of Surge Analysis with particular Reference to a Water Supply Problem. Massachusetts Institute of Technology. Paper No. Macmillan and Co'. A mer. T. Symposium on Surges in Pipelines. F. New York. 1965.. Report No. Nov. N. England.. 1960. K. Belgium.1969. Z. Nov. "Modellbeha~dlung Mllt. Mech. Massachusetts Institute of Technology. Mech. pp. Hydrodynamics Lab. 89. Rome.. "An Investigation of the Effect ofCavitanon Bubble on the Momentum Loss in Transient Pipe Flow. Mass. Boari. Th~rley. B. Basic Engineering Amer. C. Engrs. 25. A. Streeter. July 1963. of Civil Engineering. G. Bergeron. V..edung No. Mathematical Method for Digital Computers. A.. Soc. Dec. pp. pp. Eagleson. 71." Proc.. Civil Engrs. "Waterhammer March 1965. 165179. ' Swamlnathan. E. Div.. S. A. Amer.... Soc. H." Jour. March 1971. vol.. Analysis of Surge.. Sirnin.1963. New York. 1961. 79112. "Some Applications of Waterhammer Analysis by the Method of Characteristics.. Soc. ' Safwat. pp. (See also Discussion by Goldberg. 1969. on B~umeister. 27. Abbott. of Mcch. J. 19591971. Third Edition.). M.. 21." Thesis presented to the UmvefSlty of Ghent. R. Water Works Assoc Pressure Pipe. 24." Proc." Master's Thesis. Sw~ffield.4. Civil Engrs. L. Amer.. Engrs. 4th ed." Jour. Fluid Engineering. "The Velocity of Waterhammer Waves. London." Jour. B.. R. Second Conf on Pressure Surges. G. Allievi. Nov. Pickford.. D. 1963. pp. in Incompressible pp ." Jour. Ghent. New York.. Engrs. Sept. New York.. Mech. 33. June 1977. 1956. A. 59.. Ricardo Garoni. ." Proc. 10th International Assoc. W. R. 455460.. Halliwell. Inc." unne s. 19. 97105.. Engrs..pp. R. Proc. vol. J. P.. John Wiley & Sons.. 1964.8 for Waterhammer pp. J.. Technical University ' R~hm. Part I No. 20. "Measurements of Transient Flow Velocities 1m~r. V.. Soc. New York. 1970." Proc. Parmakian. S. 1904. P.. Kennison. V. Thorley. 1937." in Raison. 10. Roark. pp. and Larsen. "~ber die Druckstossgcschwindigkeit in Rohrleitungen mit quadratischen und rech teckigcn Querschnitten. McGrawHili Book Co. Rich. 13. "Hydropower Plant Transients. No. Standard Handbook for Mechanical Engineers. Applications" ' Tnkha. 117121. 29 196869 Wood. (eds.. Amer." Jour. II. L.. "Propagation of Transient Pressure Waves in ~ SodlUmC??led Fast Reactor. Evangelisti.42 Applied Hydraulic Transients Equations of Unsteady Flow Through Closed Conduits 43 6. L. L. A. G. pp.pp. C.. H. "A Laboratory Investigation of Transient Pressure Waves In PreStressed Concrete Pipes. "Theory of Waterhammer. No. published by British Hydromechanic Research Assoc. 102.. 68WA/FE2. Streeter. published by British Hydromechanic Research Association. and Rossi..pp.pp. Amer. "Water Hammer in PVC and Reinforced Plastic Pipe.. "Permanen~e e~ nietpermanente stromingen door leidingen in Kunststof. pp.. Nos. Enever. H. pp. R. 8. ' We:Ier. Thorl:y.) 28. R. "Pressure SUrge Propagation in ThickWalled Con. Meeh.. Amer. vol. S. Dover Publications. Guerrini. and Funk.:. 7. Manual of Engineering Hyd. K:. R. Amer.. Jenkncr. M.. G.99103. Sept. A mer. Streeter.. Soc. vol. 112. . Jour. A mer. Hydraulics Div... 1967. 98. Part 3E. Evangelisti.. 341424. July 1976. 1965.309324. First International Conference on Pressure Surges. 29.. 110. ~. Series D vol. An Introduction to the Method of Characteristics.. Mech. Nov." Symp... J. Engrs. Civil Engrs. A. A. 1977. A. G. pp. 25. 1976. 1970.4753. W. A mer. McGrawHili Book Co. Soc. .. 1973. Nov. vol.831843. HY4. Aug. 759770. American Elsevier. J. Streeter. Research. Institution ofMech. London. . "Velocity of a Waterhammer Wave in an Elastic Pipe. G. 1967.8 pp. 13231328. pp.. Div. Sept. 180. McGrawHIlI Book Co. vol.. Practice No. H... Formulas for Stress and Strain. M. Committee July 1942... 9.. Hirose.'ent Method for Simulating FrequencyDependent Friction in Transient Liquid Flow. Watters.. "'" water Power. Amer.. Basic Engineering. ' . "Pressure Transients in Hydraulics Paper No. " McGrawHill Book Ne w Y k or." fur Wasserbau and Wasserwirtschaft. L.703713. F. Institution 603614. Inc. Amer.. No. 839858. 7th ed. W. Civil Engrs. 31. Lister. of~ech. "Waterhammer Analysis by the Method of Characteristics. S. pp. Wave Velocity in C onere tTl e Pipelines... duits of Rectangular Cross Section. and Wilf. D. V. England." translated by E. vol. AllAIIl. May 1971. Advanced Engineering Mathematics. on Pressure Transients. ADDITIONAL REFERENCES Translated by O. 1. May 1962. Soc. D. . L. and Flammer.. Wylie. and Lindvall. K. R. Halmos.. and Wylie. A. "The Application of Heavisides Operational Calculus to the Solution of Problems in Waterhammer . "Waterhammer Analysis.pp. Jeppson." Jour. 56. Part 3E. M. HY6. HY3. F.An Effi~. R.. 1012. Pearsall. 71FE29. John Wiley & Sons. M. M. Soc. 1969. and Twyman.. 1966. 8397.. A. "SurgeWave VelocitYConcrete Engrs.. Civil Engineers. and Guyrner . T. 16. 180.." Trans." Proc. D..1965. 24. P. 1967.. Hyd. 25. pp. Eng. 14. G. "Waterhammer Analysis Including Fluid Friction. Cranfield." L 'Energia Elettrica.. London.. G. E. J. Dept. New York. 15. "The Numerical Solution of Hyperbolic Partial Differential Equations by the Method of Characteristics. vol. Canterbury. Tison. The City University. Waterhammer Analysis.. H. J. 23." Schweizerische Bauzeitung vol 89 Feb 1971 pp." Part II and III. 673692. Engrs.. London. 183. 12. E. Engrs.. 1971. "The Influence PIpe Flow. Mech. 32.
30) derived in the last chapter as dQ _ gA dH f dt .1) (3.4) and dH :::aH + aH dx dt at ax dt By defining the unknown multiplier (3.a' dt (3.1 ~.ave con.10) dx :::a dt (3.::: A.11) Let us consider a linear combination of Eqs.9) 3.. Readers having an elementary knowledge of partial differential equations should be able to follow the development of these equations. we will follow the general approach proposed by Lister!' and later adopted by Streeter and Wylie. partial differential equations.11 IS satisfied.gA (aH + at I f . In other words. The equations for simulating a conduit are derived.(if + 2DA QIQI:::0 (3. t) and Q ::: Q(x.11 and 2. let us rewrite the dynamic (Eqs.i 1[1: EQUATIONS and continuity equations if (3.4 and 3. 3. 1 through 9.9 ~nd 3. ax 2DA =0 (3. The stability and convergence criteria for the stability of the finitedifference scheme are then presented. The chapter concludes with the presentation of a case study..5. Note .10 is valid jfEq. those interested in a rigorous treatment should refer to Refs.CHAPTER 3 or Method of Characteristics 45 METHOD OF CHARACTERISTICS (aQ a t + A. then the total dQ _ aQ aQ dx + . L :::LI + AL2 44 .3 can be written dQ + gA dH + _f__ _ dt a dt 2DA QIQI. by imposing the relations given by Eqs.e.8) Ii Ii dx dt (3.1 and 3.lO A number of innovations presented by Evangellsti? will also be outlined. We will endeavor to keep the derivation of the equations free of advanced mathematics. or A.7) and by using Eqs.3) If ~ :::. 3.5) A. Eq. we h. and a procedure for the analysis of piping systems is outlined. 2. 3..verted the partial differential equations (Eqs.aH) +QIQI A.0 if :::a and (3. In deriving these equations.1 3. i. The details of the method of characteristics are presented in this chapter. 3.8 is valid if Eq. In the last dt at ax dt (3. it was demonstrated that the equations describing the flow in closed conduits are hyperbolic. 3.2 CHARACTERISTIC 1J.that ~q: 3.a2 a aQ) x + A. and a number of methods available for their solution were discussed.2.:::± 1 a as (3.1 and 3.2.1 INTRODUCTION chapter. and the boundary conditions for a number of simple end conditions are developed. as dx :::A. t) are solutions derivatives may be written as of Eqs.1 and 3. 3.9 is satisfied and that Eq.2) To facilitate discussion. 3.2) Into ordinary differential equations in the independent variable t.H(x. 3. 3.6) transientstate 1.
i.10) are valid along the pipe length (l. there is a constanthead reservoir at the upper end (at x = 0) and a valve at the downstream end (at x = L). 3.9 and 3. To facilitate discussion..l( p Inilial B Condilions t= 0 ~_'_' A x=o = B x =L ot. 3. 3. It is clear from this figure that the conditions in Region I depend only upon the initial conditions because the upstream boundary conditions did not change. these lines divide the xt plane into two regions. a pressure wave travels in the upstream direction.10) 1{7 t. it will be represented by line BC as shown in Fig. let us first discuss the physical Significance of characteristic lines in the xt plane. Because of this pressure rise.8 and 3.10.3). For example. If excitations are imposed simultaneously at points A and B. 3.2.3. Thus. they represent the path traversed by a disturbance.1) at time tv would reach point P after time !:::." Physically.5. In other words. x . for 0 <x <L) and special boundary conditions are required at the ends (i.4. 3. which may be dominated by two different kinds of solution.2. the characteristic line AC separates the regions Reservior Valve Flow L Figure 3.e. If the path of this wave is plotted on the xt plane. 3. 3. 3. Mathematically.t.8 and 3. In the example under consideration. 3. the characteristic line BC separates the two types of solutions.11 represent two straight lines having slopes ±l/a. then the region influenced by the initial conditions is as shown in Fig. Region of of Validity Equations Compalibilily ( Eqs. Eqs. Regions of validity for a single pipeline. Prior to presenting a procedure for solving Eqs. and the line BC separates the regions influenced by the downstream boundary and the initial conditions. a disturbance at point A (Fig..e. These are called characteristic lines. whereas in Region II they depend upon the conditions imposed by the downstream boundary. The compatibility equations (Eqs. the solution may be discontinuous along these lines. t.t x Figure 3. Let us assume that there is steady flow in the pipe at time t = 0 when the valve is instantaneously closed. Figure 3. influenced by the upstream boundary and the initial conditions. let us consider a single pipeline shown in Fig. This reduces the flow through the valve to zero and results in a pressure rise at the valve.8 and 3. at x = 0 and at x = L) (Fig. and the transient conditions are produced by closing the valve.e.. the characteristic lines on the xt plane represent the traveling paths of perturbations initiated at various locations in the system. Single pipeline.46 Applied Hydraulic Transients Method of Characteristics 47 In the xt plane. Characteristic lines in xI plane.
Note that Eq.8 through 3. dQ=QpdH=HpHB Region I QB (3.e.12 through 3.48 Applied Hydraulic Transients dH=HpHA I = . For such cases.14) (3.23) .2DA QBIQBI (3.18 is valid along the positive characteristic line AP and Eq..12) (3. 3.. 3. a predictorcorrector method or a secondorder approximation (see Section 7.1 Method of Characteristics 49 (3. 3. we can write along the positive characteristic line AP. 3. then a firstorder approximation may yield unstable results. we can write along the negative characteristic line BP. namely. 3.19 along the negative characteristic line BP. these are initial steadystate conditions) or have been calculated for the previous time step. Evangelisti'' suggests a predictorcorrector method.15 refer to the locations on the xt plane. we obtain x Figure 3.18) and Eq. However.1 .13) Similarly. Referring to Fig. 3.17 as (3.12 and 3. computing the friction term at the points A and B.HB and (3.20) (3.QA + . These are either initially known (i.Hp and Qp...1.19) in which _ gA ttlt Cp .4) should be used to avoid instability of the finitedifference scheme.15) 1=0 A B The subscripts in Eqs. and Lister!' employs both first. let the conditions at time t = to be known.16) and To solve Eqs.S.16 can be written as (3. a firstorder technique suggested by Streeter and Wylie is sufficiently accurate and is discussed here. The values of these unknowns can be determined by simultaneously solving these equations. if the friction losses are large. 3. (3. (3.18 as the positive characteristic equation and Eq. 3. Excitation at downstream end.22) ttlt . Because the time intervals used in solving these equations for practical problems are usually small.15 into Eq. 3. We will refer to Eq.2DA QAIQAI gA Cn = QB .e. a number of finitedifference schemes have been proposed: Streeter and WylielO use a firstorder finitedifference technique. and the constant Ca 'depends upon the conduit properties. (3.8 and Eqs.13 into Eq. The values of the constants Cp and Cn are known for each time step.10.HA . Excitation at upstream and downstream ends. 3. In Eqs. 3. and multiplying throughout by tlt.19 as the negative characteristic equation. 3.19. We want to compute the unknown conditions at to + tlt.21) I = AI c t =0 x Figure 3.11.18 and 3.. 3. we have two unknowns.4. at t = 0.14 and 3.17) Equation 3.0. 3. i.and secondorder finitedifference schemes. Substituting Eqs. Referring to Fig.
18.2.~a. ConstantHead Reservoir at Upstream End (Fig. Equation i. These are developed by solving Eq. The pipeline is divided into n equal reaches (Fig. then n. as discussed above. 3.6). 3. Eqs. Now conditions at t = to + t.18 and 3. =Hres (3.19 for the upstream boundaries. Characteristic grid. 3. and special boundary conditions are used for the end conditions.6 shows that the conditions at the boundaries at t = to + M must be known for calculating the conditions at t = to + 2t. Then. '~I~ c£nergy Grode Line \:rlI_I_~ __ 'Hydraulic Grode Line Flow Oalum Line (a) Flow Datum Line (b) Figure 3. by using Eqs. Equation 3.18 or 3. Constantlevel upstream reservoir.t.3 BOUNDARY CONDITIONS In the last section we discussed that special boundary conditions are required to determine the conditions at the boundaries.6) at the end of the time step can be determined.t at the interior points adjacent to the boundaries. Therefore. in which Hres = height of the reservoir water surface above the datum. and the steadystate conditions at the grid points at t = to are first obtained.26) • Now the value of Hp can be determined either from Eq. 3. 3.18 and 3. . 3.24) o A In/erior sections Downs/ream boundary Ups/ream boundary Figure 3. and the conditions at t = to + 2t.. the computations proceed stepbystep until transient conditions for the required time are determined.7.t are known at all the grid points. In this manner. 3. 3. to determine the conditions at t = to + t.2gA2 (3. in Chapter 10. 3. at the boundaries. he . we will again consider the single pipeline of Fig. 3. conditions at all interior points (see Fig. 3.18 is used for the downstream boundaries and Eq. and the conditions imposed by the boundary.18 or Eq.t are determined by following the procedure just outlined. 3.25) However. To illustrate how to use the above equations.t. However. Thus.7) x If the entrance losses as well as the velocity head are negligible.23 are used for the interior points.50 Applied Hydraulic Transients t Method of Characteristics 51 3. if the velocity head or the entrance losses are not small.6.23. A number of simple boundary conditions are developed in this section while complex boundary conditions such as for pumps and turbines are derived in Chapter 4 and 5 and for waterhammer control devices. then these may be considered in the analysis as follows: Let the entrance losses be given by the equation _ kQ'j.19 is available. 3.19. 3. '.19 for the upper end thus becomes (3. A close look at Fig. or both. either Eq.19. we need special boundary conditions to determine the condition at the boundaries at time to + t.
3. . Qp = O. HydrauliC Grade Line Flow x .8) If the head losses at the entrance to the reservoir are h e k . 3. 3. 3. (3. =0 (3.CaHres } (3. 3..31) Solving Eq.27 simultaneously. k is assigned a negative value for the reverse flow.7.k)P_ Q2 2gA 2 (3.8a (3.Ca(l .27.34) kQ2 =_p_ Now Hp may be determined from Eq. 3. Figure 3..9) N_~ o<LJ~ At the dead end. Qp _ 1 +v'1  4k2k3 2k2 (3.29) Now H p can be determined from Eq. then (3. 3.p=H res Method of Characteristics 53 in which k is the coefficient of entrance loss. 3.35) 2gA2 Energy Grade Line '"8"~"'m« I'" N Qp = Cp ..37) Flow Datum Line (a) ::r: ~ . it follows that (3.19) H. 3. 3.. Constantlevel downstream reservoir.18.(1. from the positive characteristic equation (Eq..18). For the reverse flow. 3. Elimination of Hp from Eqs.9.27 and 3. Referring to Fig.30.52 Applied Hydraulic Transients then referring to Fig.27) equation (Eq. Hence.18 yields k2 Q}  Qp + k. ____ O_e_a_d_E.18 that If the exit loss and the velocity . Hp=Hres(l+k) Q}2 2gA .30) and it follows from Eq.CaHres a.32) in which in which (3. 3. ConstantHead Reservoir at Downstream End (Fig.8. and the negative characteristic In Eq.k) 2 2gA2 and k3 = Cp ._nd_~ Datum Line ( b) Figure 3.32 for Qp.. Dead end. Energy Grade Line"" (3. .36) ~~ Dead End at Downstream End (Fig. 3. 3. head are negligible.31 and 3. k is assigned a negative value in Egs.29.« '"o I''" "'" .33) Solving Eq.
. taking square of both sides and defining the relative valve opening 7 versus t curves (Fig.38._. The first subscript will designate the conduit number. 3. . indicates flow at the jth section of the ith conduit. 3. 3.1 Ob and c) may be specified either in a tabular form or by an algebraic expression. we obtain Orifice at Lower End Q~ Substitution 3.' L:~. Valve at downstream end.39) Dividing Eq.11) Hydraulic Grade Line (a) In the preceding discussion. 3.18. . ':.40 yields = (%7)2 Hp o (3. To compute the transientstate conditions in which subscript a indicates steadystate conditions. 3.40) equation (Eq.11. and A v area of the valve opening.. However. the subscript P will indicate the unknown variables at the end of the time step. no special care had to be taken to designate the variables at the boundary since there was only one conduit section under consideration. This simplifies presentation and at the same time does not result in any ambiguity since each conduit can have only one endsection at a boundary. As discussed previously.. 3. the opening may be used with T = 1. Therefore. For variables that have sa~e value at all sections of a conduit.. 3. For example.42) Now Hp may be determined from Eq. For this purpose.10) a valve can be written Qo = (CdAv)o. remains constant. ' i' j. For an orifice. Although Cp and Cn may have different values at different sections of a conduit.~_. Therefore. T (3. only one subscript will be used.38) <. Method of Characteristics Solving for Qp and neglecting the negative 55 sign Valve at Downstream Steadystate in which Cv = (TQo)2/(CaHo). only one subscript will be used with them to 'indicate the conduit number. if the boundary is at the junction of two or more conduits. the above equations for Hp from the positive characteristic Series Junction (3. Conduit I (b) Opening (c) Closing  i I Conduit  i +1 i. we will use two subscripts. while the second will indicate the section number. 3. "'. Qp ..1 Figure 3.38 may be written for the transient state as = = for an opening or a closing valve.~J" 54 Applied Hydraulic Transients End (Fig. 3. we considered only one conduit.. with the radical term flow through (3./2gHo as (3.41 ) (Fig. Figure 3_10.39 by Eq. Note that T = 1 corresponds to a valve opening at which the flow through the valve is Qo under a head of H o = (CdAv){(CdAv)o.22) for the ith conduit. Series junction. H 0 head upstream of the valve. For example..n+l~'i+I. Cd = coefficient of discharge. then the variables at different sections of various conduits have to be specified. and the boundary was either at the upstream or at the downstream end. An equation similar to Eq.~.18) into Eq.. Cai refers to constant Ca (Eq.
46.44 through (3.n+l = Qpi+l.56 Applied Hydraulic Transients Now 3.44) (3.l Qk· 1+~. t+ Method of Characteristics 1 I' Hp.50) Figure 3. n + 1) and (i + 1. Equation for total head (3.1 h 12 2gAj+1 Hp.52) (347) . Continuity equation Qpi. .n+l = Cnj+1 + Cai+IHPj+l. and it is assumed that the velocity heads in all conduits are equal. 3. +(I+k) Pj+l. 1 can be determined from Eqs.l QPj+2.l. 1 from Eqs.12) For the branching junction shown in Fig.n+ I' 57 If the difference in the velocity heads at sections (i. 1) are QPj. if the difference in the velocity heads at sections (i.55 and 3.1 (3.1 .  CQjHpj.44 through 3.56) In Eqs.n+l The positive and negative characteristic equations for sections (i.1 (3. 1+1.54) 3.43 through 3.Qp.l (3.z+ 1. hi. 3.55 and 3..54.46 that 2. 3.n+l =Hpj+1.45) = Cni+1 + The continuity equation at the junction is QPj. '+2.12. QPj+l. 1 can be determined from Eqs. 1) (Fig. 3.12.55.. the head losses at the junction are neglected. . .1 and Hp.43 through 3. '+1.49) in which (3. I.48 yields QPj. In such cases. 3.n+l Caj+1HPi+l.56. 1) or the head losses at the junction are not negligible.1 2gAj+l (3.43 is not valid. 3. However.n+l and Hp. Simultaneous solution of Eqs. at the junction kQ" Pj+l.l + QPj+2.Jb2 .n+ l' and Qp.48) in which k = coefficient of head losses. the following equations can be written: I. 11 r+ . and 3. 3. Simultaneous solution of Eqs.46) = QPj+l. n + 1) and (i + 1.l = Cni+2 + CQj+2Hpj+2. the following equation for the total head may be used instead of Eq. and and Qp.n+l Qpi+l. '. Qp.+ 1 1 I can be determined from Eqs. 3.1 = b + .4cd 2c (3.45.57) Now Qp. Characteristic equations H Pj.43) Branching Junction (Fig..+2. 3. 3. 1 (3. n + 1) and (i + 1. C +C Qj 1 c.51 through 3.n+! + Q~.l = c.11) and the head losses at the junction are neglected.3.46.53) (3. 3.1 (3. Branching junction.1 1+ aj+l Now Hp.55 yields (3.43: H Pj. then Hpj. 3. 3. III+l=H 2gA. then Eq.56.Qp.52 through 3. .51) It follows from Eqs.1I+1 = Cp.
P'!" Methods for determining the convergence or stability criteria for nonlinear equations are extremely difficult. 1 + Vi + 4CaCs(Cn + CaC7) Qp = 2C C as Now Hp can be determined from Eq. Using the procedure proposed by O'Brien et a1. it is necessary that the same time increment be used for all conduits so that boundary conditions at the junction may be used.64) is satisfied. 3.61 ) The finitedifference scheme presented in Section 3. If the nonlinear terms are relatively small. St . This time increment should be selected such that Courant's stability condition (Eq. it is reasonable to assume that the criteria applicable to the simplified equations are also valid for the original nonlinear equations.60. 3. Headdischarge curve for a centrifugal pump.19. 3. This curve can be approximated by the equation Hp=C7C8Q~ (3. In other words..14 and considering the linearized equations.e. Francis Turbine at Downstream End The headdischarge curve for a Francis turbine running at constant speed (i. 3. The convergence and stability may. 3.59) llt llx a (3. If the time interval.58 Applied Hydraulic Transients Method of Characteristics 59 Hp Qp Figure 3.18) yields (3. Perkins et a1.16 and 3.63) The criteria for convergence indicate that the most accurate solutions are obtained if Eq. It has been proved that convergence implies stability and that stability implies convergence. the convergence and/or stability criterion for the finitedifference equations (Eqs. 3.17) is given by the expression ~~l_ llx "" a (3.4 STABILITY AND CONVERGENCE CONDITIONS I + VI + 4CaCIO(Cp  CaC9) (3. then llx must be greater than allt to satisfy Courant's stability criteria. = Solving this equation simultaneously with Eq. For a neutral scheme. Centrifugal Pump at Upstream End The headdischarge curve for a centrifugal pump running at constant speed is shown in Fig. however.13. the scheme is stable. the characteristics through P pass through R Qp=~~2CaCIO Now Hp can be determined from Eq. 3. 3. 3. if not impossible. Thus.2 to be stable.62) This condition implies that the characteristics through point P in Fig.60) Solving this equation simultaneously with the positive characteristic equation (Eq.13.64) This is called Courant's stability condition.58.9 showed that for the finitedifference scheme of Section 3. connected to a large system) and at constant gate opening can be approximated as (3.63 is satisfied.58) differential equations as llt and llx approach zero. is such that the reach length for any conduit in the system is not equal to allt.2 is termed convergent if the exact solution of the difference equations approaches that of the original . 3. If the roundoff error due to representation of the irrational numbers by a finite number of significant digits grows as the solution progresses. (3. the scheme is called unstable. be st udied analytically by linearizing the basic equations. Collatz'? suggests that the convergence and stability may be studied by numerically solving the equations for a number of Sx] llt ratios and then examining the results. if this error decays.5 SELECTION OF TIME INCREMENT FOR A COMPLEX PIPING SYSTEM For a complex system of two or more conduits.1 should not fall outside the segment AB. 3.
however. must be an integer and is equal to the ith conduit is divided.6 COMBINED IMPLICITCHARACTERISTIC METHOD L· ail1i (i = 1 to N) (3.!? propose a procedure called zooming in which At for longer conduits may be integral multiples of At for shorter conduits of the system. Courant's stability condition. The validity of this technique is questionable20 because the original governing equations are arbitrarily altered. and N = number of pipes velocity is not precisely known. In the author's opinion. 3.19 and floods in rivers. Different values of a may be used for different conduits. a time step equal to exAt is permissible.1 S. computed at the grid points only while conditions at Rand S must be known to determine conditions at P. Yow's technique. (Hi+1 + Hi) 2At (3. 3. . thus. For the analysis of such systems.66) =' aH at . As the wave in its value are acceptable. At = ' tion is multiplied by an arbitrary factor a2• The resulting equation and the continuity equation are then converted into the characteristic form. To avoid this.60 Applied Hydraulic Transients Method of Characteristics 61 and S and not through the grid points A and B (Fig. is applicable only to those systems in which the inertial term is small as compared to the friction term such as gas flow in pipes/S. Yow'" has reported a technique that allows larger time steps and at the same time satisfies the Courant's condition. However. the derivatives of the continuity and dynamic equations (Eqs. 3. The conditions at every time step are. Notation for implicit method. however.14. Let us consider a piping system in which the ith reach of a conduit is to be analyzed using the implicit method. otherwise.14). In this method. the implicit method combined with the characteristic method should be used if a number of conduits in the system are very short relative to others. and the value of a may be as large as 20. simple adjustment of wave velocities to satisfy the following equation should give sufficiently accurate results. the inertial term of the dynamic equa aH ax = (HPi+1 + Hi+l)  (Hpi + Hi) 2Ax (Hp'+l + Hp.65) in which n.)  (3. it was pointed out that sometimes it is advantageous to use a combination of the implicit and characteristic methods while analyzing certain piping systems.2) are replaced by the centeredimplicit finite differences!" as follows (Fig. Because of multiplication by a 2. whereas Kaplan et al. later investigations have shown that this procedure smooths the sharp transient peaks. 3. l+l x Figure 3. In the last section. Details of this follow. B.15): a large amount of computer time is required for analyzing systems having very slowly varying transients. in which At is the time step given by the Courant's condition. In this technique. and C. Notation for interpolation.67) 8 AI( x Figure 3.1 and 3.21 flow in porous media. extreme caution must be exercised while using this technique for the analysis of these systems. minor adjustments Because of the limitations imposed on At by the number of reaches into which in the system. Streeter 16 suggests that the original differential equations for short conduits may be written in an implicit form. Streeter and Lai in their pioneer paperlS and Streeter and Wylie 10 proposed an interpolation procedure for computing conditions at Rand S from the known conditions at A.
5(Qi+l + Qi) To simplify presentation.19) and Hpi+! :::: res are the other H two equations.44.. i.69) (3.72. namely.62 Applied Hydraulic Transients aQ :::: Method of Characteristics 63 (Qpi+l + + Qi+l) . similar equations are written. Pipe No.18 and at the boundaries from the appropriate boundary conditions.3.1 and 3. 3.40 were used to determine the conditions at the valve.65. 3. Seven points on the rt curve were stored in the computer. In the author's opinion. are provided by the end conditions of the reach.72) in which gAAt Cll =Ax (3. the variables.43. The steadystate discharge and pressure head at all the sections are then computed. For example.71 and 3. is obtained.73) To compute transientstate conditions in a piping system. there should be four eq~~tions. QPi' QPi+l' and Hp For a unique solution of these equations. and the transient conditions are computed at all the interior points from Eqs. there are four equations in four unknowns.71 and 3.5. should give sufficiently accurate results. The time is now incremented. 3. and the conditions at the interior sections were determined using Eqs.i a time interval of to 2~ of the transit time. the shortest conduit in the system is divided into a number of reaches so that a desired computational time interval.17 b.(Qi+l = aQ + Qi) (3.65 or the procedure outlined in Section 3. care should be taken that its end conditions are simple. Equations 3.25 s.25) were used to determine the conditions at the upstream end.17a were determined. Therefore.75) (3. h (3. 3. 3. Substitution fication of the resulting I only one subscript is used in this section to designate of Eqs. and At should be increased or decreased depending upon the rate at which transients are produced.70) 2At Q = O. This process is continued until transient conditions for the required time are computed.23 and 3. Conditions at t = At at all sections of the system were now known. this criterion should be used as a rough guide. in addition to Eqs. iih~se other two equations.(QPi + Qi) ax at (3. This procedure was repeated until transients for the desired duration were computed. . 3.72.3.68) 3. Time was incremented by At.76) Note that there are four unknowns in Eqs.1 and No. At. 3.2. however. then the negative characteristic equation (Eq.47 were used to determine conditions at the junction of pipes No. 3.t+ 1 .24 and 3.23 and 3. As the valveclosure time is rather large as compared to the wavetransit time in the system. Thus.6 is used so that characteristics pass through the grid points.42 and 3. and their values are printed. wavetravel time from one end of the system to the other. The boundary conditions for the reservoir (Eqs. The conditions were printed every second time step by specifying IPRINT = 2.66 to 3.71) (3. pipe No.2 and simpliyield J I equations Q».3. and their values are determined by simultaneously solving these equations.1 was also divided into two reaches to satisfy Eq. According to Evangelisti. 3. thus giving At = 0. Note that the conditions imposed by the boundary are used for the additional equations and not the boundary conditions developed in Section 3.18. 3.7 ANALYSIS OF A PIPING SYSTEM 2Ax (Qpi+l QPi) . 3. These were stored as conditions at the beginning of the next time step. To illustrate this procedure. Hp The flowchart of Fig. the remaining conduits in the system are divided into reaches having equal lengths by using the procedure outlined in Section 3. while selecting the conduit or the conduit reach for which the implicit method is being used. Transient conditions were caused by closing the valve according to the rt curve shown in Fig.e. 3.46. and 3. If necessary.CIlHp. + CIIHp'+l + C12 ::::0 (3. + Qp. and Eqs. Having selected the value of At. For this purpose. 3. the computer program of Appendix B was developed in FORTRAN IV language. and the r values at the intermediate times were parabolically interpolated. 3. and the initial steadystate conditions were computed.2 was divided into two reaches.70 into Eqs. the wave velocities are adjusted to satisfy Eq. if there is a conduit at the upstream end and a constanthead reservoir at the downstream end.16 shows the computational steps for determining the transient conditions in a series piping system. transient conditions in the piping system shown in Fig. For any other end conditions.
It is a peaking power plant.1/sec L = 550m D = O.OIOm L 450m D = O.16. Series piping system.l 0. Flowchart for a series piping system.2 2 Time 3 4 5 6 (seconds) ( b) Valve closure curve Figure 3.17.64 Applied Hydraulic Transients Method of Characteristics Iniliol steady stole Hydraulic grade line 65 Reservoir Valve Pipe No.23 located in British Columbia. Figure 3.75m a = IIOOm/s f = O.S 0.8 CASE STUDY Figure 3. and owned by the British Columbia Hydro and Power Authority.0 o. Canada.18 shows the schematic layout of the conduits of the Jordan River Redevelopment22.6 Io. .2 I m.OI2m (0) Piping system 1.6m a = 900m/s f = O.4 0.1 Qo: Pipe Na. 3.
and a 1400mIong penstock reducing in diameter from 3." . I/) '0 e u Q IU UJ I/) I/) Ul !=I ..19) were stored in the computer at 20 percent intervals of the valve stroke.5m rated head.78) in which 0 z III a: :::> I .J Z ~ :::> C!I UJ~ a:E C:::> UJE Q: . To reduce the maximum transientstate pressures.77 are simultaneously solved...~ g:i o z « UJ vided.19.. in which A = crosssectional area of the conduit just upstream of the PRY.5 m as determined from the prototype tests.J U ! a . To develop the boundary condition for the PRY..18.. Note that both H. and the discharge at the intermediate valve openings was determined by linear interpolation. Noting that Qp = Qu and eliminating H n froru these equations. There is only one Francis turbine rated at 154 MW and 265. both at valve opening T.!E 01/)0: • 0: ~.. 3. Hn = H p + Q~/(2gA 2). Points on the PRY rating curve (Fig. « o l z > (3.18 and 3.* Analysis of transients caused by various turbine operations is discussed in Chapter S. 3. a computer program was developed by using the boundary conditions for the PRY derived in this section. 3. c I/) z !! . i.2 to 2..) of 265. *Boundary conditions for the simultaneous operation of the PRY and wicket gates are developed in Section 10.66 Applied Hydraulic Transients Method of Characteristics 67 The upstream conduit consists of a tunnel having a 5285mlong. Eqs. To determine the transient conditions caused by opening or closing of PRY.7.. 3. e: u e: 0 a: .7 m. . at rated head (H. 82mlong.2mdiameter sections. 3. .96mdiameter. and Qr = discharge under rated net head Hr. The rating curve for the PRY. is shown in Fig. and Hn are total heads. and 451mIong.. mainly Dshaped section.77) in which Qu = PRY discharge under a net head of Hn.e. 3. the PRY discharge under net head Hn is given by the equation (3. a pressureregulating valve (PRY) is proUJ ~ ~ C!I . Assuming that the valve characteristics obtained under steadystate operation are valid during the transient state." . (3.s::.79) U Now Hp may be determined from Eq.
. '" . Thus. ~ ~ . The friction factor for various conduits were computed such that they included the friction and minor losses. contraction. and assuming the penstock to be anchored at the lower end and free for longitudinal expansion at the upper end. .20.. The natural frequency of the cell was greater than 1000 Hz and it was calibrated against the deadweight gauge. which delivered linear output within 0. these are assumed to be distributed along the conduit length. In the computer analysis. the PRY was first opened from 0 to 20 percent rololype 20 Gole 40 opening 60 (%) 80 100 Figure 3.._ .19. Comparison of computed and measured results.. and bend losses.. I II gaugetype pressure cell. Lined and unlined segments of the tunnel were combined into two lined and unlined reaches. the upstream conduit was represented by 11 pipes while the conduit downstream of the PRY was neglected because of its short length.. In the author's opinion. . and a Westinghouse leadingedge flowrneter+" was used to measure the transientstate flows. In the prototype test. . The waterhammer wave velocity''" was computed by taking the modulus of rigidity of the rock as 5.35 m... <:I..6 percent over its entire range. this approximation should not introduce large errors in the analysis.. 3.. A number of transientstate tests were conducted on the prototype. The head losses computed using these values of friction factors and those measured on the prototype are in close agreement. and the Dshaped tunnel was replaced by a circular conduit having the same crosssectional area.!:! 10 o o V t/ tl I. <> /~ Model. '" ~ <::) . such as expansion. Transientstate pressures were measured with a strain 'b ~ 340 . ~ 30 20 ..68 APplied Hydraulic Transients 70 T ! Method of Characteristics 69 60 'b <: .y/ 50 40 I:: <> s .. A multiturn potentiometer mechanically connected to the PRYstroke mechanism was used to measure the PRY opening."' t ~ ~ II.!:.. Discharge characteristics of pressureregulating valve. The computed and measured transientstate pressures and flows are shown in Fig. although the minor losses are concentrated at various locations in the actual system. I I If ~.24 GPa. Steadystate pressures were measured by a Budenberg deadweight gauge having a certified accuracy of 0.: 330 320 310 I I I 2 Time (Stlc) Figure 3_20.
3.4. 3. 3.10) is not neglected and there is an additional term V(aH/ax) in the continuity equation. 3. 1012. 3. 4. Is the equation for QPj n+1 given in Problem 3. NewYork. Develop the boundary conditions for a centrifugal pump running at rated speed.2 is 90mIong instead of 450 m as shown in the figure. 1950. For illustration purposes. A procedure called Zooming is presented in Ref.. John . pp. SUMMARY in which C = Q~/(CaAHo) is the orifice head loss for In this chapter.I = C and AHo + "'. 3.17a. Run the program for various values of At and plot a graph between the computed pressure at the valve and At."L 'Energia Elettrica. 1969.7.). there is good agreement between the shapes of the pressure curves but the measured results show that the pressure waves are dissipated more rapidly than that indicated by the results of the mathematical model. Stability and convergence conditions for the finitedifference scheme were discussed. Stoker.H. Develop the boundary conditions for an opening or closing valve located at the junction of two conduits (Fig.5 valid for the reverse flow? If not.18. (Hint: The following four equations are available: the positive characteristic equation for section i.9 Valve Conduit i I Conduit i +1 i. "WaterhammerAnalysisby the Method of Characteristics. Inc. (eds. (Hint: Solve the system using the zooming procedure and then using the same At for the whole system as determined by the Courant's condition. Dover Publications. A. the negative characteristic equation for section i + I. G. 3. the computed and measured transient pressures agree closely up to an elapsed time of about 18 s.G. As can be seen from Fig.21 is replaced by an orifice and the conduits i and i + I have the same diameter. These differences may be due to using the steadystate friction formula for computing the transientstate friction losses and the reduction of wave velocity at low pressures as discussed in Section 2. S. Water Waves. however.. Assume that pipe No. and wall material. and Wilf. Evangelisti. then QPj. 3. Write a computer program for the piping system shown in Fig. 3. 2. tion. and a number of simple boundary conditions were developed. The PRY was then closed from 20 percent to 0 (Fig..9.n+1 = QPj+1 . the PRY was not completely closed but was held at 1 percent opening to simulate the leakage through the wicket gates. 1. 3. 759770. and a procedure was outlined for the selection of time interval for a complex system. Valveat seriesjunction.. However.A.20.7. Mathematical Methods for Digital Computers. lnterscience.3. 1960. Prepare a flowchart for programming the boundary conditions developed in Problem 3. Ralston.6.C 2 + C(Cp + Cn) Qo. Solve these equations simultaneously to obtain an expression for Qp. Develop the boundary conditions for the pressureregulating valve and the Francis turbine shown in Fig. The transient conditions are caused by opening or closing the valve. taking into consideration transients in the suction line. derive a similar e'quation for the reverse flow. 17 in which the time step for the long pipes may be an integral multiple of that for short pipes.8.6. the continuity equa REFERENCES Wiley& Sons.2. In the computer analysis.NewYork. 839858. Prove that if the valve in Fig. 2. and the equation for flow through the valve. Nos..21). 3.n+l . Prove that the equations of the characteristic curves are dxldt = V ± a if the term V(a Vlax) of the dynamic equation (Eq. 673692. The wicket gates were kept closed throughout the test. 3. a computational procedure for analyzing transient conditions caused by closing a valve in a series system was presented. 3.5. J.) PROBLEMS 3. wall thickness. The chapter was concluded by comparing the computed and measured results for the transient conditions caused by the closure of a pressureregulating valve in a hydroelectric generating station. NewYork. 3.70 Applied Hydraulic Transients Method of Characteristics 71 at a very slow rate and was kept at this opening until steady flow was established in the upstream conduit.1. 3. 1965. The computed and measured discharge agree closely. Webster.21. the measured period of the pressure oscillations is less than the computed period. Assume that the turbine speed and the wicketgate opening remain constant during the transientstate conditions. the procedure requires extrapolation at the junction of pipes having different time steps. afterward. Investigate the effect of extrapolation on the pressure peaks for the piping system shown in Fig. 3.) 3.17a. 1.20).Inc. J.l. n + I.n+I/~i+I' 1 Figure 3. In addition. the details of the method of characteristics were presented. Partial Differential Equations of Mathematical Physics.
1968. published by the University of Alberta.. "Computer Analysis of Waterhammer in Power Station Cooling Water Systems. P. May 1971. Engrs.1967. Eagleson. 42.175178J '. Proceedings. June 1976. of Mech. Symposium. Contractor. E. Inc. 119... 68WAI Aut. E457E472. pp. V. Paper 7365. 1968. June 1965. Soc." Amer. A433A438. Thud Berlin. Canterbury. Engrs. England." Jour. and Henson. Soc. Nov. Wylie." chapter in Ref 1. . V. Amer. Part II. Engrs.72 Applied Hydraulic Transients to the Method of Compressible of the Dissipation of Characteristics. of Civil Engrs. Fox. R. A. M.. Academic Press.. . vol. 1972. A. Series D. Yow. vol. 1966. 3543. Hyman. Mises. "A Quasi Method of Characteristics with Application to Fluid Lines with Frequency Dependent Wall Shear and Heat Transfer. Lister. pp. H. of Mech. M. Institution of Civil Engrs. 18. Kaplan. Interscience. eport o. Civil Engrs. Nov. 17. and Lai. J. 94. T. McGrawHili r: .. Virginia Polytechnic Institute. .. 223251. A. C. pp. 21. and Portfors. Hyd. 22. G. Sept. "On the Numerical Solution of the Equations of Propagation by the Method of Characteristics. Div.. Streeter. E. published by British Hydrornechanics Research Assoc. H. published by British Hydrornechanic Research Assoc. Canada. L. Edmonton. F. Methods of Mathematical Physics. 9. vol. 20. Engrs. Feb. 8. 165179. 1969. Soc. Div. G... Lai. "UnsteadyState Calculations in Complex Pipe Systems. 23. 71WAFE13. B. C. Soc. Math.. Massachussetts Institute of Technology. Waterhammer in Pumped Storage Projects. Sept... Paper No. 10." Amer.. 7. Series D. V. Chicago. D. 24. "Waterhammer Analysis. (See also Discussion by Stoner. Soc. First International Conference on Pressure Surges. vol.. M. NcwcastleuponTyne. 1972... L. of Petroleum Natural Gas Transient Engrs. A. pp. E. "The Reflection of Waterhammer Pressure Waves from Minor Losses. pp. pp. V." Internat. pp. Springer.. Boston." hoc. Sept. Mech. L. L. Streeter. published by British Hydromechanics Research Assoc. pp. Hydrodynamics La. 1960. A.. of Petroleum Engrs. and Kaplan.No... Sept. S. vol. 1972." hoc. H. Symposium on Unsteady Flow in Open Channels.. D. "The Westinghouse Leading Edge Ultrasonic Flow Measurement System. and Wylie. pp. pp... Engrs. Chaudhry. vol. H. Paper No.. "Economic Methods for Modelling Hydraulic Transient Simulation. "Method of Characteristics Applied to Calculation of Surge Tank Oscillations. E .. G. pp. "Transient Aquifer Flows by Characteristics Method. E. 14. and Stoner. Hydraulic Transients. An Introduction Theory Analysis ADDITIONAL REFERENCES New York. 12.. B. and Ippen. 1/2. of Mech. First International Conference on Pressure Surges.. 102. No. Wylie. c.. B. R. "The Numerical Solution of Hyperbolic Partial Differential the Method of Characteristics." Trans. vol. 79112.. 25. April 1974. Miyashiro. Dec." Jour. W. "The Solution of Hyperbolic Partial Differential Equations Methods. York. Amer. of Civil Engineering... no. pp. 15. Parmakian. Fox. Streeter... N. 49. New York. Martin. First Canadian Hydraulic Conference. Soc. 1.. Fischer. F. 1966.. of Civ.. 1962." Proc." Proc. 293305. "A Mathematical Model for Analyzing Hydraulic Transients in a Hydroelectric Power Plant." Ref 17. 1974. and Todd. Perkins." Jour.pp. 73 5. pp. HbydRropoweNr Transients. Hyd. Fluid Flow.. The Numerical Treatment of Differential Equations. vol. . 1. 1971. First International Con!erence on Pressure Surges. Japan Soc. C. A. 1972. A. of Energy Waterhammer. 298314. M... "The Use of the Digital Computer in the Solution of Waterhammer Problems. pp. Portfors. 29. C. In Method of Characteristics American Elsevier. Soc. pp. Engrs... 127131. 1965. Hyd. P. by Difference ...pp. E. Amer. V. . Brown. "The Prediction of the Magnitudes of Pressure Transients Generated by a Train Entering a Single Tunnel. Soc. 422428. T. 11761194. of Civ. . Soc." Meccanica. 1963. and Authors' Reply. A. .. 165175. 1967." chapter in Ref 1.9 pp. DI1DI16. 2936. Soc. G. and Zaoui... 1964.." Trans. 1951. and Wentworth.. Evangelisti.. ... 11. May 1973. B. 2nd International Conference on Numerical Methods in Geomechanics. EIIEI12. . Rachford. R. Wylie." Proc. March 1976. M. Belonogoff. Canterbury England.. 16. . Book Co. L.. 6. May 1973. Paper 7020.. b Equations y . DIV.. . England. B. 87. and Physics. "Some Computational Aspects of One and Two Dimensional Unsteady now Simulation by the Method of Characteristics. Engrs.1967... 10. April 1976.. Courant. J. Streeter.. Amer. A.~. . E. 19591971. "Waterharnmer Analysis of Pump System. M. J. H. "A Fast.. Mathematical York. 39. "Unsteady Flow Calculations by Numerical Methods. G. r 22. New . published by British Hydromechanics Research Assoc. "OneDimensional Soil Transients by Characteristics." Proc." Jour." Jour. 13. O'Brien. Dover Publications. 952958. . of Mech." presented at the spring meeting. Amer.. Highly Accurate Means of Modeling Transient Flow in Gas Pipeline Systems by Variational Methods. May." Proc. pp.. 5369." La Houille Blanche. and Chaudhry. Collatz. Sept. L. 1972. S. "Waterhammer Analysis Including Fluid Friction. J." Bull. 2. G. Soc. "Analysis and Prototype Verification of Hydrau IC Transients in Jordan River Power Plant. Combes. "Numerical Error on Natural Gas Transient Calculations. 1953. 2. 195202. M... L. Gray. ed. 1958. pp. Virginia. T. Waterhammer Analysis.. Dept.. M. No. of Mech. M. "A Study of the Numerical Solution of Partial Differential Equations.. A mer. . S. S. New " P. Sheer. of Mech." Jour. Engrs. D. No. Jan.7. V. Streeter. Abbott. 19. "The . published by Amer.. C. pp. Tedrow. Fox. New York. Soc. England. and Streeter. Engrs. published by British Hydromechanics Research Assoc. roc. Amer. "Analyse des erreurs introduites par l'utilisation pratique de la methode des caracteristiques dans le calcul des coups de belier . A. 1962. vol. First International Conference on Pressure Surges.Tnstitution of Civil Engrs. Nov.
remedial measures should be taken. For this purpose. 4 and 8. The discharge of a centrifugal pump depends upon the rotational speed. T.e. and liquid entrained in the pump impeller. then procedures outlined in this chapter should be used for the transient analysis. Various authors have presented these curves in different graphical forms suitable for graphical<" or computer58 analysis.6 appears to be the most suitable and is used herein.3 MATHEMATICAL REPRESENTATION OF A PUMP As discussed in Chapter 3. During the design stages. i. This will be discussed in detail in Chapter 9. respectively. Because of the reverse flow. motor. Transients caused by both of these operations may be analyzed by using the boundary conditions developed in Chapter 3 since the pump speed remains almost constant during the transients in the piping system. and then the power supply to the pump motor is switched off. With the increase in the reverse speed. special boundary conditions for the pump end of a pipeline have to be developed. the pump speed reduces since the pump inertia is usually small compared to that of the liquid in the discharge line. vacuum pressure may occur. Usually. and the transientstate speed changes depend upon torque. sudden power failure) are usually severe. stops momentarily.. Transients caused by emergency pump operations (e. the pump is said to be operating in the zone of energy dissipation.1 INTRODUCTION The starting or stopping of pumps causes transients in pumping installations. negative pressure waves propagate downstream in the discharge line. and Thave to be specified for the mathematical representation of a pump. and the pipeline should be designed to withstand positive and 74 negative pressures caused by these operations. 4. in a normal pumpstopping procedure. and the pressure head. Thus. In this chapter. and the chapter concludes by a presentation of a case study. the relationship between the discharge. the valve is gradually opened. Because the flow and the pumping head at the pump are reduced.. when there is reverse flow through the pump while it is rotating in the normal direction). A procedure for storing the pump characteristics in a digital computer is outlined. except that presented in Refs. then the characteristics of a pump having about the same specific speed may be used as an approximation. the method of characteristics presented in Chapter 3 may be used. H. and positive and negative pressure waves are produced in the discharge and suction lines. H. the discharge valve is usually kept closed to reduce the electrical load on the pump motor. The curves showing the relationships between these variables are called the pump characteristics. the discharge valve is first closed slowly. and then reverses. Q. Excessive pressure will be produced when the two columns later rejoin. and a typical problem is solved. at the boundary must be known in order to develop the boundary conditions. 4. If the complete characteristics data are not available. If the pipeline profile is such that the transientstate hydraulic grade line falls below the pipeline at any point. Q. the method used by Marchal et a1. little data. if necessary. if the pumps are not started or stopped as previously outlined. four variablesnamely.g.2 TRANSIENT CONDITIONS CAUSED BY VARIOUS PUMP OPERATIONS During a pump startup. are available for either the zone of energy dissipation or the zone of turbine operation. Design criteria for designing pipelines are then presented. and. However. Following a power failure. To analyze these transients. Of all the methods proposed for storing pump characteristics in a digital computer. the possibility of watercolumn separation should be investigated. In this condition (i. Although pumpcharacteristics data in the pumping zone are usually available. the analysis of transients caused by various pump operations is presented. H. N. transientstate speed changes have to be taken into consideration in the analysis. and the combined moment of inertia of the pump. N. Since the pumping head and flow depend upon the pump speed. Flow in the discharge line reduces rapidly to zero and then reverses through the pump even though the latter may still be rotating in the normal direction.CHAPTER 4 Transients Caused by Centrifugal Pumps 75 TRANSIENTS CAUSED BY CENTRIFUGAL PUMPS 4. and as the pump speed reaches the rated speed. the reverse flow through the pump is reduced due to choking effect. the pump is now operating as a turbine. and the pumping head.. and the water column in the pipeline may separate at that point. and positive pressure waves propagate upstream in the suction line. The pump speed increases in the reverse direction until it reaches the runaway speed. .e. boundary conditions for a pump end are developed. the pump slows down rapidly.
and HR is in ft.500 gpm units. suggest that sgn (h)Jlhl/(a2 + u2) be used to increase accuracy for smaller values of this parameter (sgn designates sign of h). For homologous pumps. A124 and A127).5) N T (4.1 and in Appendix E. 4.P characteristic curves for pumps having specific speed** of 25.3) NR TR ~= In this equation. H N2 and = Constant (4. the characteristic equation (equations if the boundary has pipes on both the upstream and downstream sides) and the conditions im*MarchaJ et al. * The signs of v and a depend upon the zones of operation. For a doublesuction pump. 8.4) v == Constant \r:"~ . T == torque and the subscript R designates the value of the variables for the rated conditions. NR is in rpm. and 261 SI units (1276. i. In Sl units.JQR/HR 3/4.4). Using the data presented by Thomas.4 BOUNDARY CONDITIONS FOR PUMP FAILURE As discussed in Chapter 3. 4. (J H N2D2 and N QD3 + (J" 90 0 0 == Constant + 0 0 180 + 270 (J .1) == 0 . . **Specific speed = NR. the torque characteristic curve may' be plotted between (3/(cx2 + v2) and ().1. h/a2 becomes infinite. To avoid this. On the basis ofEq. 180 o . QR is in m 3/s. in gpm units.. respectively) are presented in Fig. Similar to the pressurehead curve. 77 Data for prototype pump characteristics are obtained from model test results by using homologous relationships.2 may be nondimensionalized by using the quantities for the rated condition as reference values.4.. the following ratios are valid Zone of Operation Pump Energy dissipation Turbine Turbine energy dissipation u a + Range of 0 90 0 . NR is in rpm. and HR is in m. Transients Caused by Centrifugal Pumps Zones of pump operation.7600.. a new variable (J may be defined" as (J == tan"! ~ Q = Constant Equation 4. +Sorne authors erroneously use a specific speed of 35 SI units for this pump.e. Since D is constant for a particular unit. Let us define the following dimensionless variables: v=QR N Q H h==HR a= and then the characteristic curve may be plotted between e and h/(a2 + v2).76 Applied Hydraulic Transients TABLE 4. As the pump had a double suction. e .. and its value varies between 00 and 3600 for the four zones of operation (see Table 4. 270 0 (4. However. v (4. h/(a2 + u2) is used herein because it simplifies the derivation of the boundary conditions for the pump end (see Section 4. p. olv becomes infinite for v == O. 4. it may be included in the constants of Eq. the parameter h/(a2 + v2) instead of h/a2 may be used.nt} ni} (4... 360 0 Constant in which D = diameter of impeller. 4.1. e is always finite. QR is in gprn. To avoid this." Two pumps (or turbines) are considered homologous if they are geometrically similar and the streamflow pattern through them is also similar. and 13. Eq. By definition.t 147. In addition to the need to define a different characteristic curve for each zone of operation.3 may be written as ~=  con". rated discharge should be divided by two to compute the specific speed (see Closure of Ref. QR is divided by 2 while computing the specific speed. .2.2) Since a becomes zero while analyzing transients for all four zones of operation.1)...
For a pump end.. V Approximation of pump characteristic curves by segmented straight lines....r Yil.. For any value of (X and v (except when both (X and v are simultaneously zero)..../ /.lRt~(~~ I )" t '. }.... we have to simultaneously solve these equations to develop the boundary conditions for the pump end .. ~.78 Applied Hydraulic Transients Ii Transients Caused by Centrifugal Pumps 79 ]. . this function computes the value of fJ between 0 and 7T and between 0 0 .. \ (a) Pressure " ". To facilitate understanding of the derivation.IJ I I I.\ :!..2... 120 tonI ': 160 . If a sufficient number of points (e.. (J II ~ . ...~.' \ \ Ns = .. '\ 147 (\I \\ zoo .. ___. let us first consider a simple system having only one pump and a very short suction line. ". Characteristics of pumps of various specific speeds.. Thus. the pump characteristics define the conditions imposed by the boundary. r \ . However. (b) / I I .l!S. . eo I r" / yNs=261..~. 73) are stored.s:. .) Equations of Conditions Imposed by Pump 24"'. \\ .g.. 4. i ~ posed by the boundary are solved simultaneously to determine the boundary conditions.... . Figure 4.2). the value of (J = tan " «(X/v) may be determined by using IBM function ATAN2. We will develop the boundary conditions for more complex cases in the next section. To use these curves in a mathematical model.\.. ..in deQrees 200 " 'I. 1. Actual curv« ~ 270 120 !60 (} • IonI e: Approximating straight Iin6 Torqutl 8 = tanI~ Figure 4.. Each segment of these curves between the points stored in the computer may be approximated by straight lines (Fig. in which fJ = tan " «(X/v). then the error introduced by approximating the curves by segmental straight lines is negligible. \ \ 210 2.\ • \ \ As we outlined in Section 4..~ (. pump characteristics may be mathematically represented by curves between fJ and h/(a? + v2) and between fJ and ~/(a2 + v2). ..rj ~[L"'...IJt. 2... ~I~ I.'" eo (J. between the range (J = 0 and (J = 360 are stored in the computer. discrete points on these curves at equal intervals of (J.1.3.. + (\I > .. / . u.h~ [fJ'. and a differential equation defines the variation of the pump speed with time following power failure.
their values determined by extrapolation from the known values for the previous time steps. the constants" for the equation of the segmental straight line are determined. then the value of 0 to be used for determining the point on the pump characteristic curve is 360 . is small.10) in which Cv = coefficient of head losses in the valve. as a first estimate.. Hp = pumping head at the end of the time step.9) and Y2).X I)' (X)'YI) .I Differential Equation of Rotating Masses The accelerating torque for a rotational system is equal to the product of the angular acceleration and the polar moment of inertia of the system. 4.8) (jp I ap vp ap + Vp in which al and a. that the variables a. then 22 U :0 :J: Q.e.b..30 = 330°. the preceding linear extrapolation should yield sufficiently accurate estimates if the size of the cornputational time step. which is usually small. Since the pump speed and the pump discharge vary gradually.th time step. Notation of boundary conditions for pump.7 to 4. I IQp. and fj at the beginning of this time step are known.Y2x J )/(X2 .vi_1 are the variation of these variables during the (i .. and fjp. Referring to Fig. :J: .1. From these. I.I)th time step. and b. e. since the values of these variables are initially unknown. the decelerat = Hsuc + Hp . Equations 4. is not taken into consideration in Eq. Now.X J) and 02 = (Y2 .Y J )/(x2 . (4.9.1 I.Hp u (4. Q~. the following equation can be written for the total head at the pump: Hp. Oischar9~ yo/vtl Pump Datum Figure 4. Note that the velocity head in the discharge pipe. v. Let us assume that the calculation has progressed to the .7) (4. assuming that the points corresponding to ap.3. This limitation can be circumvented by adding 21T to the computed value of e if e < 0. To determine the value of these variables.t. hp ap +vp 2 2 = al + a2 tan = a3 + Q4 tan _I ap vp (4. Now. the grid points on either side of 0 = tan " (ae/ve) are searched. The valve head loss is given by the equation: (4. and a3 and a4 are constants for the straight lines representing the head and torque characteristics. if 0 given by this function is . Transients Caused by Centrifugal Pumps > 81 and 1T. vp. ai and Vi refer to known values at the beginning of the ith time step. i.hp. hp.30°. 1. respectively.6) in which ae and Ve are the estimated values at the end of ith time step. vp..10 represent the conditions imposed by the boundary. b.Hpv = head loss in the discharge valve. is written as Qp. II to account for the reverse flow.3. Y =01 +02X . we have to first of all determine the equation of the segment of pump characteristics corresponding to ap and Vp.1 *If (X2. in which Hsuc = height of the liquid surface in the suction reservoir above datum.ai_1 and b. 4. Let us denote these unknown variables by ap.80 Applied Hydraulic Transients :J: <l Q.g. Since there is no external torque acting on the pump following power failure. h. and that we want to compute the values of these variables at the end of the time step. and the ordinates h/(a2 + v2) and fj/(a2 + v2) for these grid points are determined from the stored values. whereas our range of interest is between 0 and 21T. we may use. and b. and (jp lie on these straight lines. then 0I is the equation of a straight line passing through the points = (Y JX2 . However. Note that in this equation.
By using an average value of (3 during the time step. Cip and Vp. 4. and 4. NR in both SI and English units is in rpm. Vp.12 have to be divided by the acceleration of gravity. (4.22)** v~) = v~) + avp "Superscript (1) indicates estimated values and superscript (2) indicates values after first iteration.19) (4. and TR is in lbft.18in four unknownscos..! I. 4.10. the righthand side of Eq.l. Therefore. 4.20 is Ci~) = Ci~) + OCip (4. in the English units. shaft.12) + CaHsuc + CaHRhp . and liquid entrained in the pump impeller.8. which may be taken equal to Cie and Ve as determined from Eq.14. 4.4. 4. and wand N are rotational speed of the pump. in rad/s and in rpm. and T/R = pump efficiency at rated conditions..20 are nonlinear equations in two unknowns. 4. motor. respectively. (4. T=WR or 2 1 dw Transients Caused by Centrifugal Pumps Continuity Equation Since there is no storage between the suction reservoir and section (i.8 into Eq. WR2 is in kg m 2 and TR is in Nm.19 and 4.17 simultaneously. TR = 60 rHRQR/(27TNRT/R) in which r = specific weight of liquid. which is then refined to a required degree of accuracy by successive iterations. .7 into Eq. we have to solve Eqs.3.18 and for {3p from Eq.21)** (4. the resulting equation may be written as QRvp = Cn = . These equations can be solved by using the NewtonRaphson method in which a solution of the equations is first guessed.4. By eliminating Hp.11 may be written as: (3 Solution of Governing Equations To develop the boundary conditions. we need only the characteristic equation for the discharge line. 4.11 and 4. it may be neglected in the analysis.15 6 TRAt rrWR2NR (4.20) Equations 4.11) in which WR2 = combined polar moment of inertia of the pump. the righthand sides. By substituting for hp from Eq.* Then.14) which may be simplified to in which c .16.15)* Characteristic Equation for Discharge Pipe As the suction line is short.17) = flow through the pump at the end of the time step. tlHp v .15 has to be multiplied by g.e. and Qp. "'''These equations may be deduced from the general derivation presented on p.e.17 and by using QR and HR as reference values. 4. {3p.l .18) In this equation. we obtain Cip . we will first eliminate hp and {3p from these equations as discussed in the following. this equation may be written in a finitedifference form as: = Now we have four equationsi.7 through 4.CaCvQh vplvpl (4.82 Applied Hydraulic Transients ing torque is the pump torque.16) "In English units. a better estimate of the solution of Eqs. In SI units. 4. g. In English units.7.14 and simplifying. 4. WR2 is in Ibft2. and 4.WR2 2rrNR dCi 60 TR dt (4. Hence. of Eqs. To simplify the solution. i. 1) Qp.16. (4. Eqs. On the basis of Eq.Ci tlt 60 TR 27TWR2NR (4. Let Ci~) and v~) be the initially estimated values of solution. 4. 4. and 4. 4.! in which Qp 83 dt = Qp (4. 4. hp. 4.9. 4.13) (4. 4.6..10. 91.19 and 4. 1).for section ti. from Eqs.14. Eq.
Transients Caused by Centrifugal Pumps 84 Applied Hydraulic Transients 85 in which F2oap = of) oUp oF2 oUp _ F) OF2 aup aFt oF2 oUp aap OF2 ) aap sr. then 0l~2) and are solutions of Eqs.23 and 4. 4. and the abovementioned procedure is repeated. Ol~) and u~) are assumed equal to 0l~2) and U~2). Having determined Olp and up. a counter may be used so that the computations are stopped if the number of iterations exceeds a specified value (e.19 and 4. Hp and Qp from Eq.24.e. Otherwise.1 from Eqs.26) OlP) up OF2 a Up =C6 ( 2a3up+a40lp2a4uptan t (4.7 and 4. illustrates this procedure. and Fl and their derivatives with respect to ap and Up are evaluated at a~) and u~t). then hp and ~p are determined from Eqs.00]). Flowchart for boundary conditions for pump.1. u~) Figure 4. 4.28) OlP) up If IOOlpl and loupl are less than a specified tolerance (e.3.23) oUp = oFt F F 20ap aFt aup aF2_ aap (4.4. 4.9 and 4.17.8. and Hp.aap aup sr.27) (4. and the above procedure is repeated until a solution is obtained. functions F. Differentiation of Eqs. and the solution progresses to the next time step. it is verified whether the segment of the pump characteristic used in the computations corresponds to Olp and up. 30).1. 0.. 4.4. To avoid an unlimited number of iterations in the case of divergence of solution.25) (4. a = Olp and ~ = ~p). The flowchart of Fig..20 yields the following expressions for these derivatives: NO YES ap) Up OlP) QR Up (4.g. 4. 4. However.1 and Qp. If it does not. oap (4.19 and 4. then Ole and Ve are assumed equal to Olp and up..20. if the correct segment was used.24) In Eqs. 4. aF2 .g. The values of a and u are initialized for the next time step (i. .
86
Applied Hydraulic Transients
Transients Caused by Centrifugal Pumps
87
4.5 BOUNDARY CONDITIONS FOR SPECIAL CASES In Section 4.4, boundary conditions were developed for a system having only one pump and a short suction line. Because of the short length, the prop~gation of the waterhammer waves in the suction line was neglected. In this section, we will develop boundary conditions for complex systems often found in practice. Boundary conditions for systems not covered herein may be developed by following a similar procedure. We will briefly describe the system configuration, and then present the governing equations and the expressions for F1, F2, aFI/aOip, aFI taup, aF2 taOip, and aF2 taup. Using these expressions, the solutions may be determined as outlined in Section 4.4. Parallel Pumps Systems having parallel pumps to which power fails simultaneously may be analyzed as follows: If the length of pipe between each pump and the discharge manifold is long, then each pump may be handled as outlined in Section 4.4, and the parallel piping system may be analyzed using the boundary conditions presented in Chapter 3 (note that the discharge manifold will be considered as a junction of two or more pipes). However, if the pipe between each pump and the discharge manifold is short, then this pipe may be neglected in the analysis, and the combined discharge of all pumps may be considered as the flow at the upstream side of the discharge manifold. Boundary conditions for the latter case are developed in this section. The continuity equation for this case is: (4.29) in which np = number of parallel pumps. Depending upon the length of the suction line, boundary conditions for parallel pumps may be divided into the following two cases:
(4.32) Expression for F2, aFl /aOip, aF2 /aOip, and aF2 tavp are given by Eqs. 4.20, 4.25, 4.27, and 4.28, respectively. 2. Long suction line (Fig. 4.5). If the suction line is not short compared to the discharge line, then waterhammer in the former has to be considered in the analysis. Therefore, we have to include the characteristic equation for the suction line. Referring to Fig. 4.5, Hp = Hpj Qpi,n+l = Qpi+I,1
QPj,n+1 + 1, 1 
Hpj,n +1
(4.33) (4.34) (4.3 5) (4.36)
c,  CaiHpi,n+l
= Cn + Cai+IHPi+I,1
= QPj+I,1 = npQp
In addition, Eqs. 4.7, 4.8, and 4.14 are valid for this case. By multiplying Eq. 4.34 by Ca.1+ I' Eq. 4.35 by Ca" substituting for Qp. l,n+l , I and Qp.1+1,1 from Eq. 4.36, and adding the resulting equations, we obtain npQp(Caj
+ Caj+l)
=
c.c; + CpCaj+1
+ Ca/"'aj+IHp
(4.3 7)
By using QR and HR as reference values, Eq. 4.37 may be written as
(4.3 8)
1. Short suction line. If the suction line is short, then the waterhammer waves in this line may be neglected. On the basis of Eq. 4.29, Eq. 4.18 becomes
npQRup
= Cn
+ CaHsuc + CaHRhp
a, ::t:
_::t:
 CaCuQk vplupl
(4.30)
Equations 4.7, 4.8, and 4.14 are valid for this case as well. Proceeding similarly as in Section 4.4, the following expressions are obtained: FI
Instantaneous grade line
hydraulic Pump
a_
+
Discharge Pipe i+1
line
= CaHRal(Oif, + uf,) + CaHRa2(0if, + uf,)
tan"! Oip  npQRvp
Up
I i+I,1 i,n+1
Datum
(4.31)
Figure 4.5.
Pump with long suction line.
88
Applied Hydraulic Transients
Transients Caused by Centrifugal Pumps
89
Elimination of hp from Eqs. 4.7 and 4.38 yields (4.39) in which C7 = C=
8
tion line. However, if the pipe length between the pumps is short, then this pipe may be neglected in the analysis, and the combined boundary conditions for both the pumping units may be developed as discussed in the following. Referring to Fig. 4.6, the following equations may be written for the system: I. Pumping head
np(Ca·
1
+ Ca·
1+
1 )QR
(4.40)
Hpj+I,1
=Hpj,n+1
+Hpb
+HPm
 flHpv
(4.44)
CajCaj+IHR CnCa·
1
2. Continuity equations (4.41) we obtain (4.42) (4.43)
F2, aF2/acxp, QPj,n+1 Qpb Qpi+I,1 +a2vp
+ CpCa· 1+ 1
CXp and vp,
I
= npQpb
= QPm
(4.45) (4.46) (4.47)
CajCai+IHR
By differentiating Eq. 4.39 with respect to
aFI a
CXp
== 2alCXP
+ 2alCXp tan
+ 2a2vp
tan

CXp vp
= npQPm
3. Positive characteristic equation for suction line
Qpi,n+1
aFI 
a~
== 2al vr
~l
cxp
~

a2cxp 
C7
= Cp
 Ca;Hp;,n+,
(4.48)
4. Negative characteristic equation for discharge line and
Qpi+I,1
Equations 4.20, 4.27, and 4.28 define the expressions for
aF2/avp.
= Cn + Cai+IHPi+I,1 = Cv Q Pi+I,1 IQ Pi+I,1 1
(4.49)
5. Equation for head loss in the valve Series Pumps (Fig. 4.6) If the pipe length between the two pumpingsets is long, then each pumpingset may be analyzed individually assuming the downstream pumps have a long sueflH Pv
(4.50)
6. Equations for the pump characteristics hp m =al
(CI'.p +Vp )+a2 mm m
2 2
~,t:i
"""",<I:L._..L_
Instantaneous grade line
hydraulic
(CI'.p +Vp2 ) tan " mm m
2
.sm. V
Pm
CXp
(4.51)
(4.52)
Cl'.p
_.!!!.
::t:
s:
...: +
fjp
m
= a3 mm (CI'.~
+ V~ m ) + a4 mm (CI'.~ + Vp2 ) tan I m + a4b
(CI'.~b + v~b)
V Pm
(4.53)
Suction
line
Main 'pump Discharge valve Pipe (i+ I) Discharge Datum Figure 4.6. Notation for series pumps. line
fjPb = a3b (CI'.~b + vh)
,
tan I
~
vPb
(4.54)
7. Equation for the rotating masses (i.e., equations similar to Eq. 4.14)
CY.Pm C6mfjPm cxPb  C6bfjPb
= Cl'.m
+ C6mfjm
(4.55) (4.56)
= Cl'.b+ C6bPb
90
Applied Hydraulic Transients
Transients Caused by Centrifugal Pumps
91
In the preceding equations, subscripts b, m, and v refer to the booster and main pump and to the valve, respectively; np = number of pumpingsets connected in parallel; and Cu = coefficient of head loss in the valve. To solve these equations, let us first reduce the number of unknowns from 13 to three as follows: Eli . tiIon 0 f H Pi,n+I' Q Pi,n+I' H Pi+I,I' Imina to 4.50 yields
Q
Pj+I,I'
(4.61)
apm, vPm' and apb'
Qp
and Hp u from Eqs. 4.44 b'
Now we have three nonlinear equations (Eqs. 4.594.61) in three unknowns, To solve these equations by the NewtonRaphson method, we have to obtain a solution of the following equations:
aFI oap ( aapm m
+ +
aFI
aVPm aF
ovp
m
+ 
aFI
aapb aF
oap
b
)(1) = Ff!)
(I)
(4.62) (4.63) (4.64)
UsingHRm, hpmHRm
HRb, and QRm as reference values, Eq. 4.57 may be written as
(
+ hpbHRb
npQR =
C
ai+1
m vPm
+
npQRm
aaPm
aF2
oap
m
a 2 vPm
aF3 aVPm
ovp
m
+
a 2 apb
aF3 aapb
oapb
)(1) = F
)(1)
2
Caj
Vp
m
aF3 oap ( aap m m
+ 
ovp
m
+ 
oap
b
= F3 (I)
By substituting expressions for hpm and hPb from Eqs. 4.51 and 4.52 into Eq. 4.58 and simplifying the resulting equation, we obtain
In these equations, the functions FI, F2, and F3, and their derivatives, are evaluated for the estimated values of a~l) , V~l) , and a~l), and a better estimate of m m b the solution is determined from the following equations:
a(2) = a(I) + oa Pm Pm Pm V(2) = vel) +OV Pm Pm Pm a(2) = a{t) Pb Pb
(4.65) (4.66) (4.67)
+ oa
Pb
As before, the superscript in the parentheses refers to the number of the iteration. The expressions for the derivatives obtained by differentiating Eqs, 4.59 through 4.61 are
(4.59) Note that in Eq. 4.59, we have replaced VPb by vPm since both are equal as the number of main and booster pumpingsets are equal. By eliminating {3Pm from Eqs. 4.53 and 4.55 and {3Pb from Eqs. 4.54 and 4.56, we obtain
F2=ap m 
(4.68)
C6 m
[a3
m
(a~
m
+v~
m
)+a4
m
(a~
m
+v~
m
)tant
pm a] vPm
(4.69)
92
Applied Hydraulic Transients (4.70) (4.71)
Transients Caused by Centrifugal Pumps
Initial steady stole Hydraulic grade line
93
oF  2
00'.
Pm
= 1
2C6
m
Q3
m
(XP m
2
C
I:)
t;;
6
m
Q4
m
(Xp
m
tan
1
O'.Pm
Ie
:J:
t. Two pumps
ReserYoir
"o
(4.72)
(4.73) (4.74) (4.75)
L .. 450m = 0.75m a = 900 m/sec. f 0.01
o
L= o= a= f=
550m 0.75m IIOOm/sec. 0.012
Qo = 0.5 m3/sec.
oF3 ~
Pm
Pump 0010
= 2C6bQ3bVPm
 2C6bQ4bVPm
tan
I
(4.76)
If IliO'.p I, IliO'.p I, and llivp I, obtained by simultaneously solving Egs. 4.62 m b m ) (2) (2) through 4.64, are less than a specified tolerance (e.g., 0.001 , then (Xpm, vPm' and are assumed equal to O'.~) , V~2) , and O'.~b)' and the above procedure is repeated m m until a solution is obtained. Then, it is verified whether the segment of pump characteristics used in the computations corresponded to O'.p and Vp. If it does not then (Xe , (Xe , and Ve are assumed equal to (Xp(2) , (Xp(2b),and V~2) , respec, m b m m m tively, and the above procedure is repeated; otherwise, the values of the remaining variables are obtained from Eqs. 4.44 through 4.56, and the solution progresses to the next time step. To avoid an unlimited number of iterations in the case of divergence of iterations, a counter should be used so that the computations are stopped if the number of iterations exceeds a specified value (e.g., 30). 4.6 EXAMPLE To illustrate the use of the above procedure, the piping system shown in Fig. 4.7 is analyzed. Initially, both pumps are operating at rated conditions, and the
O'.p(2)are b
QR HR NR WR2
= 1100 rpm = 16.85 kg  m2 per pump at rated conditions 0.84
= 0.25 m3/sec. = 60 m
solutions of Eqs. 4.59 through 4.61: otherwise, O'.~) ,
m
v~)
m
,
and
(X~lb)
Pump efficiency
Figure 4.7.
Piping system.
transientstate conditions are caused by simultaneous failure of power to both pumps. A computer program (Appendix C) was developed using the boundary conditions derived in Section 4.5 for Parallel Pumps and the flowchart shown in Fig. 4.4. The method of characteristics discussed in Chapter 3 and the boundary conditions for the reservoir and series junction were used to analyze the transient conditions in the discharge line. The waterhammer wave velocity for various sections of the discharge line was determined using the equations presented in Section 2.6. The pumpcharacteristics data for Ns = 25 Sl units (1276 gpm units) shown in Fig. 4.1 were used in the analysis. At rated discharge and rated pump speed, the pressure head at the upstream end of the discharge line would be egual to the rated head. Starting with this flow and pressure head at the upstream
94
Applied Hydraulic Transients
Transients Caused by Centrifugal
Pumps
95
end, the steadystate conditions in the discharge line were determined. Then, the power was assumed to fail, and the resulting transient conditions were computed. As the inertia of the liquid between the pump and the discharge manifold was small, the discharge of both pumps was lumped together and considered as the flow at the upstream end of the system. Computed results are presented in Appendix C. 4.7 PUMP STARTUP In some piping systems, there is no control valve downstream of the pump; therefore startup procedures outlined in Section 4.2 cannot be used. Pump startup in such installations may produce very high pressures, especially if the motor is of the induction type and is started across the line (l.e., without reducing the voltage). The transients caused by a pump startup may be analyzed by selecting a startup time, Ts, and by assuming that the pump speed increases linearly from zero to the rated speed in time Ts' The motor manufacturer can supply the time taken by the motor to reach the rated speed. The time specified by the motor manufacturer should be decreased by about 30 percent 11 to obtain a value for Ts' Since the pump speed is known (it is assumed to increase from zero to NR, in time Ts), the data for the torque characteristics and moment of inertia, WR 2, of the pumpmotor are not required for the analysis. The pumping head and the pump discharge may be computed as outlined in the following. Estimate the nondirnensional pump discharge, Ve' at the end of the time step by extrapolation from the known values of v for the previous time steps. From the pump characteristics, determine hp for the known value of ap and for the estimated value of the pump discharge, Ve' Then, Hp = hpHR' in which the sub'" script (I, I) refers to the first section on the discharge line just downstream of the pump. Now, using this value of Hp, " compute the discharge at section (1, I) from the negative characteristic equation (Eq. 3.19), Qp 1,1
=
The pressure rise during a startup may be reduced by having a slow startup. This can be done by increasing the WR 2 of the pump motor, by reducing voltage, or by having a partwinding start. The overall economy of decreasing the maximum pressure to reduce the pipewall thickness by these methods should be investigated prior to their selection.
4.8 DESIGN CRITERIA FOR PIPELlNES Once the layout and dimensions of a piping system have been selected, the maximum and minimum pressures for various operating conditions can be determined by using the procedures outlined in Sections 4.4, 4.5, and 4.7. In the safest design, all components of the system would be designed for the possible maximum and minimum pressures with a liberal factor of safety. Such a design would, however, be uneconomical. Therefore, a factor of safety is chosen depending upon the risks and the probability of occurrence of a particular operating condition during the life of the project, i.e., the higher the probability of occurrence, the higher is the factor of safety. Based upon the frequency of occurrence, various operating conditions may be classified as normal, emergency, or catastrophic. A discussion of the operating conditions included in each of these categories and the recommended factors of safetyl2 follows.
NonnaI All those operations that are likely to occur several times during the life of the pumping system are termed normal. Appurtenances or devices (e.g., surge tanks, surge suppressors, and air valves) provided in the system to reduce severe transients are assumed to be properly designed and to function as designed during these operations. The following are considered to be normal operating conditions: I. Automatic or manual starting or tripping of pumps throughout the entire range of pumping head. If there is more than one pump on the line, all are tripped Simultaneously; however, only one may be started at a time. 2. If a check valve is present near the pump, it closes instantly upon flow reversal. 3. A surge tank does not drain and thus admit air into the pipeline, and it does not overflow unless an overflow spillway is provided. 4. If there is an air chamber, it is assumed to have a minimum air volume during a power failure.
Cn +CaHp
1,1
(4.77)
and then determine Vp = Qp j(npQR) in which np = number of parallel pumps. ',I If Ivp  Ve I, e, in which e is a specified tolerance (e.g., 0.001), proceed to the next time step. Otherwise, assume ve equal to the mean of the computed value of Vp and the estimated value, Ve, during the previous iteration, and repeat the procedure. If the discharge line is under a static head prior to the pump startup, then there will be no flow into the discharge line until the pumping head exceeds this static head. This condition can be included in the above analysis by assuming that Qp 1,1 = 0.0 until Hp 1,1 exceeds the static head.
State of California.96 Applied Hydraulic Transients Transients Caused by Centrifugal PUmps 97 As a result of any of the above operations. However. This valve opens as soon as power fails to the pumpmotors. Because the probability of occurrence of any of these conditions is extremely remote. I 3) and two large units (Nos. Mathematical Model A computer program was developed based on the boundary conditions for the pump end derived in Sections 4. are inoperative. if present in the system. and mathematical model are then briefly described. then special devices will be provided to minimize the transient pressures when the columns subsequently rejoin. The minimum and maximum total pumping heads are 159. There is a discharge valve on the downstream side of each pumping unit. boundary conditions developed in Sections 4. respectively. Sacramento. These conditions include: 1. 4.9 VERIFICATION OF MATHEMATICAL MODEL The transientstate prototype test data obtained on the Wind Gap Pumping Plant by the Mechanical Design Unit of the Department of Water Resources. such as loss of all air in the air chamber. A factor of safety of three* based on the ultimate bursting strength of the member and a suitable factor of safety against collapse are recommended for the transient pressures caused by normal operations. surge tanks. surge tanks.06 and 160.and multipleunit tests were conducted on both the small and large units. This is followed by a comparison of the computed and measured results. was used to verify a mathematical model based on the * A factor of safety of four is recommended in Ref. with the dischargevalve closure delayed until after the units had reached the steadystate runaway speed.5. 2. A valve position transducer (displacement) and an rpm (analog) transducer were installed on all units tested to record the dischargevalve closure and the unit speed. the tests.4. and is rated at 17. plant data are first presented. if it is impractical or too costly. The straingaugetype pressure transducers were used to measure the transientstate pressures on the upstream and downstream sides of the discharge valve.7 m when operating at 360 rpm. But. . Each unit has a combined pumpmotor WR 2 of 99) 366 kg m". The closure of the discharge valve following power failure may be included in the 4.8 m. 3.4 and 4. Catastrophic Catastrophic conditions are those in which the protective equipment malfunctions in the most unfavorable manner.03 m. The large units (Nos. This valve closes in 22 ± 2 s following power failure to the unit. Closure of one of the check valves provided for shutting off return flow through the pumps is delayed and occurs at the time of maximum reverse flow. 12. The specific speed of the pump is 33. 5). instrumentation.84 m3/s at a total head of 159.5 and on the flowchart of Fig. and the friction loss in the pipeline corresponding to a flow of both units is 1. a factor of safety of two based on the ultimate bursting or collapsing strength is suggested. 4. we will only list the parameters of these units and of their discharge line. Runaway tests were conducted by subjecting the units to simulated power failure. In this section. The pipeline is 628 m long and varies in thickness from 11 to 27 mm. a factor of safety of slightly more than one. then appurtenances such as air chambers. 5) are manifolded together into a 3. etc. very rapid abnormal opening or closing of a valve or a gate. Since the probability of occurrence of these conditions is rather small. 4. may be used. Since test data for the large units only were used for verification purposes. if the watercolumn separation does occur.8lmdiameter pipe. or relief valves is inoperative. One of the surge suppressors. based on the ultimate bursting or collapsing strength. Emergency The emergency operating conditions in pumping systems are those in which one of the pressurecontrol devices malfunctions during power failure. To prevent backflow from the downstream canal. with the siphon having an air valve at its top. Tests and Instrumentation S~ngle.8 (Sf units). the water column does not separate at any point in the pipeline. should be provided to avoid it. a siphon is provided near the downstream end of the pipeline. and pumpshaft failure. Airinlet valves. Plant Data The Wind Gap Pumping Plant has five pumping units: three small units (Nos.4 and 4.
v o ':' valve closed gradually (Continued) in 22 seconds o .8. ."d.1 . \ \~ \\ /' r ~ r=: ~i=~.2 14 POWER FAILURE'" SEC r. Comparison of computed and measured results... . 300 ..: r.. are presented in this section. .d.8a.. 0. Wind Gap Pumping Plant.. the agreement between the •.... Figure 4. the computed minimum pressure is lower than the measured minimum pressure. ""(I$Id...P carried out by the Hydroelectric Design Division of British Columbia Hydro and Power Authority for the water supply consultant* during the preliminary design of the makeup and coolingwater supply system for the Hat Creek Project of the Authority.SEC ~ ~•• open . .llrwI The computed and measured results were compared for the tests simulating simultaneous power failure on Unit Nos..d e L 0.*f\"j I~~. • 00 o 1ItH. 4. The pumping stations may have several parallel pumps..._. 100 \.. L II'H~"'~ " .9 \ 0.... . 0.200 ".. •• Z2 400 0. Results for the case when the discharge valve remained open are presented in Fig./ \ ~ 200 :::. " .. and results for the case when the valve was closed are shown in Fig. the. To compute the transient conditions in the pipeline.d ::. 7 \ ~PllIIIP'PNtI. :.~~ '\ ~f\. derived in Section 3. 4.p 6P •• 4. .. if desired.6 '\ I '.. which was not simulated III the mathematical model.. <t1mpu..2 Transients Caused by Centrifugal PUmps 99 analysis.. However. and the pumps may have long or short suction lines.. 4.' o TIUE AFT[R '0 POWER •• '!IOO (a) Discharge valve remained Figure 4.. *Sandwell and Company Limited (SandweIl). •2 FAILURE ..6 \ lV v 2 c.  22 TI"[ '''TER .K\ . 4 and 5 with and without closure of the discharge valve.10 CASE STUDY The hydraulic transient studies. Canada.v VIIf \ \ V ~ b__"""r.1 •.0 ._ 200 0.. Ctllftpul.00 8 \ ~ .0 ~ 9 ~'W 1(.8 0. 300 computed and measured results is satisfactory.... were used in the program. Comparison of Computed and Measured Results r...i.8. Vancouver..3 . <. ... l! L '00 1\ \~ \~ 1\ 1. method of characteristics of Chapter 3 and the boundary conditions for the downstream and upstream reservoirs and for the series junction./ ~ ...3../ / IJ).. ._... " 400 1.7 \ I~ \. As can be seen from the figure.'N '~ __. L 1\'\. \ 0. ~ .. British Columbia. " .98 Applied Hydraulic Transients .8b.::..~"". " •• .• ". ~ •• 20 _ _ l.:::: _// 0 '00 • (b) • Discharge • = 10 I. .. ~. This difference is ~ost probably d~e to the operation of the siphon valve....
a pumping station with five pumping units at the river intake. The pump manufacturer supplied the pump characteristics for the normal zone of pump operation only. respectively. The boundary conditions and solution procedures presented in Chapters 3. motor. 102 was used to select appropriate waterhammer control devices: J(8'UJ /IDMU.J0J . Since no data were available for the other zones and since these characteristics agreed closely with those of Fig.. and 3580 rpm. 2. e "" ~I ~I §I 81 gl £1 ~I ~I 01 S3!H3W .725 and 1. The river intake would be located on the right bank of the Thompson River. and the maximum total static lift from the river intake to the plant reservoir would be 1083 rn.4 km northeast of Ashcroft.. Each booster station would have the freesurface tank on the suction side. the watersupply system (see Fig. 3 Both booster stations have threestage pumps. transients in the discharge line were analyzed neglecting the effects of transients in the suction line. 3 The average and maximum discharges would be 0. problemoriented computer programs.9) or those obtained by using other available. and a reservoir near the power plant. if necessary. To avoid errors introduced by interpolations. and entrained water in the impeller is equal to 62 kg m2. each with four pumping units and a freesurface suction tank. Selection of Control Devices The procedure outlined on p.2 (SI units). 4.9) for pumping water from the Thompson River to the plant reservoir would be comprised of an 800mmdiameter buried pipeline. simpler. two booster stations.4 m /s. 4.1 were used for all zones of operation. 670 m.60 m /s. and 10 were used to solve the characteristic form of the dynamic and continuity equations. shaft. 4. approximately 23 km long. total inertia for each unit can be increased to 420 kg m 2 without exceeding the limits set by the pump startup time.1 for Ns = 25 (SI units). and the moment of inertia of pump.100 Applied Hydraulic Transients WaterSupply System As presently planned. Because there is a freesurface tank on the suction side of each booster station. British Columbia.. wave velocities were adjusted slightly. The program was verified by comparing the computed results with those measured on a prototype (see Section 4. each rated at 0. Analysis Computer Program A computer program for analyzing the transient conditions in a pipeline caused by power failure and/or valve operation was developed. the characteristics of Fig. o IE . The specific speed of each pump is 39. so that the characteristics passed through the grid points.4. If required.NOl1111\313 .
Steadystate pressure. reverse flow through the pumps for an extended period may partially drain the pipeline at high points. until better data were available and a sensi Column Separation The following control devices would successfully prevent column separation in . the pressure nse *Pressure rise" Maximum transient state pressure . the maximum pressures at the pump could be kept below 5 percent of the rated head. at the pump following power failure was less than 5 percent of the rated head. Column separation. In addition. With these controls. Increase the WR 2 of each pump motor to 370 kg m2. With the available data. the maximum pressure rise* at the pump end could be limited to 10 percent of the rated head. Discussion The above results were obtained using assumed friction factors and assumed pump characteristics. the maximum pressure at the pump in all cases remained less than the steadystate pressure. the minimum pressures in the pipeline remain above atmospheric pressure. If necessary.2.1O percent. A single rate closure was assumed in these computations. The maximum reverse pump speed following power failure for the cases when the discharge valve remained open and when the discharge valve was closed was less than the following maximum permissible limits specified by the pump manufacturer: 130 percent of rated speed for less than 30 sand 120 percent of rated speed for longer periods.9 for the system containing suitable control equipment. both the topographic information and the data for the discharge valves were not precisely known. The provision of additional inertia at the pumps and oneway surge tanks prevented column separation. the various segments of the pipeline: Maximum Pressures The pressure rise following power failure could be reduced by slowly closing t~e pumpdischarge valves. 4. However. However.102 Applied Hydraulic Transients Transients Caused by Centrifugal Pumps 103 1. it was recommended that. . the water level in the suction reservoir had to be artificially lowered to obtain the correct downstream head for the given pump speed. . Although the maximum reverse pump speed was within the limits specified by the pump manufacturer. the pressure rise may be higher due to significant variations in the data for the system. The system was analyzed for the case of simultaneous power failure to all pumps. Similarly. the pressure rise following power failure exceeded. and provide a 4mdiameter oneway surge tank at Station 17+480 with steadystate water level in the tank at EI. Artificial lowering of the suction tank should have a negligible effect on the computed pressures in the discharge line. as discussed. 1345 (25 m above ground level). However. the pressure rise could be decreased by increaSing the dischargevalve closing time. it could be reduced to less than 5 percent by slowly closing the pumpdischarge valves. 627 (10m above the ground surface). It was assumed during the initial design of the pipeline that with appropriate control devices. Two alternatives are available: (A) Increase the WR 2 of each pump motor to 115 kg m2 . 1252 (10m above ground level). and provide a 4mdiameter oneway surge tank at the top of Elephant Ridge with the steadystate water level in tank at EJ. Maximum pressure. A difference in the friction losses could also affect the maximum and minimum pressures. . With the closing times of about 100 s. 1 to 2. thus possibly resulting in situations where column separation could occur. With these measures. the minimum pressures along the pipeline were above atmospheric pressures. Therefore. Pipeline Downstream of Booster Station No. Watercolumn separation occuredin the pipeline between Booster Stations No~ . test runs using different pump characteristics showed that variation in the pump characteristics and/or in the data for the discharge valve could substantially change the computed maximum and minimum pressures. 2. 2. Because of the higher than normal inertia.. which would result in an increase in the time period for which the pump runs in the reverse direction.2. assuming there were no control d. provide a 4mdiameter oneway surge tank at Station 1 I+ 175 with steadystate water level in the tank at El. 1 and 2 and in the pipeline downstream of Booster Station No. and (B) Increase the WR 2 of each pump motor to 390 kg m2.evices. The data for these devices are listed subsequently. the discharge valves were assumed to remain open following power failure.. significant changes in the ground topography would change the hydraulic grade line relative to the pipeline. As the design operating conditions for each pumping station were different from the specified rated conditions for the pump. Pipeline from Booster Station No. Results The maximum and minimum hydraulic grade lines following power failure are shown on Fig. With check valves located downstream of the pumps. I. Emergency Conditions As an emergency condition.
Using the pump char . Ws = weight of disk in water. Air valves should be provided at high points along the pipeline.. De~elop the bo. A check valve having no dashpot a~d having negligible bearing friction losses closesl5 according to the 4. PROBLEMS in which e = angle between the center of gravity of disk and vertical' J = mome~t of inerti.9 should be adjusted proportionately.5. a pressureregulating valve IS sometimes provided just downstream of the pump.25 p3 4. a procedure for storing the pump characteristics in a digital computer was presented. To.ent for stationary disk in moving water (function of e). and . G. Expression for these constants are B _ (AR)O. and the chapter was concluded by a presentation of a case study. 4.6.ghtcenter o.4. and the elevation of the maximum hydraulic grade line shown in Fig.cl. the maximum pressure rise at the pump end should be taken equal to 10 percent of the rated head. in which power fails to 4. investigate the effect of increasing the value of WR 2 on the maximum and minimum pressures. 4. write a computer program and investigate the effect of vanous rates of opening and closure of the pressureregulating valve. if necessary. The inertia of the pump motors could be increased by adding flywheels or by a custom design of the electric motors.valve closes.1. The alternative with increased inertia only is better from an operational point of view because the oneway surge tank is not as foolproof and in addition requires constant maintenance. equation de Jd"lt 2 Wsrsme+ _.as the power fails and is closed slowly later. In order to provide operational flexibility and ease in exchanging spare parts.flow coefficient for moving disk in still water (function of e). Develop the boundary conditions fo~ such a system. With the specified control measures. however.3. When power fails to the pump. if provided.7a Vo Is . Using the program of Problem 4. C.3 of the disk. repair. Two alternatives are available to prevent column separation in the pipeline between Booster Stations Nos. etc.wa t er h ammer wave velocity V = steady.de)2 +(G (FV)2 +Kf Kd dt Kd dt K f =0 In this chapter. The oneway surge tanks should have two pipes for water outflow.104 Applied Hydraulic Transients Transients Caused by Centrifugal Pumps lOS tivity analysis of the effects of changes in the variables affecting pressure rise was made. V = mean pipeline velocity. the safety margin could be increased.3.flow coeffi. water in the discharge hne decelera~es. This should considerably reduce the possibility of a tank becoming inoperative due to the failure of a check valve to open.11 SUMMARY acteristic o~ Appendix E. etc. and boundary conditions for a number of cases usually found in practice were developed. and F are constants. Criteria for the design of pipelines were presented.2. and develop a computer program. and g = acceleration due to gravity. '0 4. These would be helpful during filling and draining of the line and would prevent collapse of a long length of the pipeline should a break occur in the pipeline at a lower elevation. an iterative procedure for analyzing transients in piping systems caused by various pump operations was outlined. state flow velocity. Draw a flowchart for the boundary condition derived in Problem 4. At Elephant Ridge and at the summits downstream of Booster Station 2. in which a . A check valve is provided in a discharge line to prevent reverse flows t. (BV C de)2 +. and B. 4. kd . I and 2. prove that the maximum pressure at the pump does not exceed 14 the pumping head if the friction losses are greater than 0. this should be investigated in detail.hrough the pumps. r = dist~ce fro~ PiVOtto wel. Write a generalpurpose computer program to determine the transientstate pressures in a discharge line caused by power failure. would be studied during the final design. In addi tion.f gravity of disk. This valve open~ . Transients caused by the operation of these valves. reduce ~aximu~ pressures following a power failure.1. the minimum hydraulic grade line was always above the pipeline. During the final design. the minimum hydraulic grade line was less than 5 m above the pipeline. nf pumps and no pumps keep operating. it was decided that all units at both the booster stations should be identical and that each would have a WR 2 equal to 400 kg m2 • 4.t?e check . valves could be provided along the line to isolate and drain segments of the line for inspection. when better data should be available.undary conditions for a system having n parallel pumps. kf =.
" Jour. Fluid Mechanics. 11.voI... H. Hydrau ICS IV. Blackie and Sons. RT "Complete Characteristics of Centrifugal Pumps and Their Use in PredicTrans. J. TN402... .77 0. pp. ''Waterhammer Caused by Pumps. DD 108. British Columbia Hydro and Power Authority. K f and K d are listed in the followmg: ex (degrees) kf kd 0. J. Write a computer program for the check valve.. London. London.09 4. Jaeger.. H. 479484. Streeter.58 0. 75. The Netherlands.pp. L. u D' P 5. 9951006. Chaudhry.eds on 4 · ons . J . an . pp. Schnyder. pp. Streeter. . New York. G. . Miyashiro. "Pressure Surges on Starting Pumps with Empty Delivery Pipes. B. vol.... c. Vancouver.1 (Section is 9 in. Kitt re d ge.. respectively.1957. P. mer.. Joseph. pp. Dover Publications.. of Mech. Parmley. John Wiley & Sons. W. ys ems. of Civil A Engrs. ''Water Level Oscillations in a Surge Tank When Starting a Pump in a Pumped Storage Power Station. except that 0 4 8 12 16 20 24 28 32 36 40 44 Use the pipeline the diameter O.di t distance from pivot to the point of concentration ~f f ~ dA. E. Engrs. Canada. Ws 0 10. 75. Engrs. . (J = 16. Dec..! data: I = 0. of Mech. 19. C. paper presented at Internattona f ymposlum a ig: V. R.. sponsored by Amer.23 0.. M. McGrawHili Book Co.IS ~nce from pivot to point of concentration of moment of inertia of disk area. First International Conference on Pressure Surges." Ref 6.84 0. 1216112168. "Waterhammer Analysis for Pumps Installed in Series.78. Centrifugal and Axial Flow Pumps. Separate No.49 0. 1963. 2nd Edition.55 0. .. . 8.B = 0. of Civil Engrs. pp. D.40 0.e .40 0.. 59. 2. vol. b .16 0.44 0. 0.74 Ibft. Sept.28 0. Rich." Trans.~. Amer Soc ofCiv. Soc.." Bull. Kilmarnock. pp.. 1976. p = in which . Research Assoc. .pp. E.1 and 44 . Stepanoff. New York. "Waterhammer Design Criteria.62 0.. pp. 7881 3P kian J Waterhammer Analysis. 5..ll.1956... Soc. Aug.. A. H. Af disk' R = distance from pivot to center of disk.. at Canterbury.. pp. A321A332. 12. G. and r = moment Develop arm measured from disk pivot. "Pressure Surges at Large Pump Installations. p. Hydroelectric Design Division. J. Elsevier Scientific Publishing Co. · Darm~y ""Complete Pump Characteristics and the Effect of Specific Spe. 1953.. that K f . Pipeline Design for Water Engineers. 10. 3. ADDITIONAL REFERENCES and pump data given in Ex. V. and Kd are given in a tabular 4.(J and ex are 60. ArneI'. 9. and Wylie. 1951.357. W. 58. Engrs. the boundary conditions for the check valve.1307132_2.. pp.. "Complete Characteristic Circle Diagrams for Turbo Machinery. 259266. 819826.. .49 0. L. Waterhammer in Pumped Storage Projects. Miyashiro. Miyashiro. 1965. P . Hydraulic Transients. of Mech." 683689. pp." Report No. Soc. McGraw Hill Book Co. 123133.54 0. and run it ~or ~e fol~OWI.67 0. vol. vol. C = 0. 79.." Jour. . 1972. "Determination of Pump Characteristics for a Computerized Transient Analysis. 13.. Soc.56 0. 11171125. 151159.. July 1964. Stephenson. Oct. of the pipelines REFERENCES K' 1. 1959.pp.. Third Edition.. IS' n · Means ate' h Diait aI C0 mputer". Hyd. J. Parrnakian. Thomas. ArneI'. V. Power Div.. pp. mer. rans. Basic Engineering. 1965. H. organized by British Hydromechanic Research Assoc.10 + ex.49 0. Parmakian. . Pump St " Trans A mer.27 0." Trans. Swanson. Engrs.. Amsterdam. Hydraulic Transients. 466468. . Linton.Engrs. assuming form. ArneI'..07. A mer. England. of Mech. of Mech.. Jaeger.ofMech. 1937. International Assoc.685699. pp. Soc. Parrnakian. .. for Hydraulic Research.Transients Caused by Centrifugal Pumps 107 106 · Applied Hydraulic Transients 7. London.. . No. Engrs.548. T A Soc of Mech Engrs. J. 1963." Jour. Glenfield and Kennedy Ltd. L. roc. Engineering Fluid Mechanics. "Hat Creek Project: Hydraulic Transient Analysis of Makeup Cooling Water Supply System. Soc.95 14. F... "The Behaviour of Check Valves during Closure. Hydraulic Transients. 1953." Water Power. Nov... "Hydraulic Transients in Centrifugal I B Dec 1961 pp. 1967.151171. Japan Soc. vol." Proc.38 0." PI'OC.. tn«. Amer. "StartUp pressures in Short Pump Discharge Lines.6). pp. 1953. Cass. Engrs. ti:~P~iT~an'~ient Behaviour.. I. 695700. and Hamill. McGrawHili Book Co. 59. Stre~ter. 1950. 15. 1962. Engrs. New York.F= O. New York.. ''Waterhammer Analysis of Pumps in Parallel Operation. of Civil Engrs. vol.area 0. 98." Research report. 361. Soc. FI h G d Suter P "The Calculation of Water hammer Problems y 6 Marchal. Soc.71 0. "Pressure Surge Control at Tracy Pumping Plant. A mer. M. pp. February 1978. of Mech. W. vol. 4. initial steadystate. "Waterhammer Analysis of Pipelines. 168188.. England. .. 133140. July 1972. esc.7." Jour. M. "Comparison Between Calculated and Test Results on Waterhammer in Pumping Plant. 1937. and Bell." British Hydromechanic." Trans. L. 0.0 0. April 1957..! .235 Ibftsec2. july 1959. Soc. ." Proc. D.
and Kennedy. 1. International Symp. 10. 619639. such as startup. load acceptance. CHAPTER 5 HYDRAULIC TRANSIENTS IN HYDROELECTRIC POWER PLANTS 5. 94. The schematic representation of a typical hydroelectric power plant is first presented. turbogenerator. The chapter concludes with the case study of the governing stability studies carried out for a 500MW hydroelectric generating station. and the selection of generator WR2 and optimum governor settings. 5. 1965. 91." Proc. Soc. Due. Japan Soc. boundary conditions for a Francis turbine connected to a large system were derived. of Civil Engrs. May 1965. In this chapter. Various turbine operations that produce hydraulic transients in the water passages of a power plant are discussed.." Jour.. "Waterhammer Charts for Centrifugal Pump Systems. Soc. Soc. of Mech.1 INTRODUCTION In Chapter 3.. 1968. An electrical generator is mechanically coupled to the turbine. F. "Negative Pressure Phenomena in Pump Pipelines. A governor is provided to correct any changes in the system frequency by opening or closing the wicket gates of the turbine. pp. Outflow from the turbine is carried downstream through the downstream conduit system. vol. Applied Hydraulic Transients H. HY3.. and the electrical output of the generator is carried by the transmission lines to the load centers. Brown. pp. Dec.. pp. vol. A mer.2 SCHEMATIC OF A HYDROELECTRIC POWER PLANT Figure 5.1967.1 shows the schematic diagram of a typical hydropower plant. lake. Engrs. HY3. of Mech." Proc.. As shown in the figure. Engrs. followed by a discussion of the governing stability of hydroturbines. Waterhammer in Pumped Storage Projects. or load rejection.. Amer. R. H. and governor are then outlined. A mer. of Mech. J.. Miyashiro. "Waterhammer Control in Centrifugal Pump Systems.. 521531. 952958. J. 247270. H. or canal. 154157. Kinno. Details of the mathematical simulation of the conduit system. Basic Engineering. ''WaterColumn Separation at Two Pumping Plants. of Civil Engrs. Prototype test results used to verify the mathematical model are then presented. vol.. 109 ." Proc. pp." Bull. a mathematical model is developed for analyzing hydraulic transients caused by various turbine operations. such as a reservoir. ''Waterhammer Analysis of Pump System. to the turbine. Engrs. Soc. 1968. pp.108 Kinno. Amer. the upstream conduits convey water from the upstream source.
the flow through a Francis turbine depends upon the net head.? this is a valid assumption. In the expressions of Table 5. or a downstream reservoir.1.1. Usually very little data are available for small wicketgate openings. steadystate model test results are used to plot the expected prototype turbine characteristics and these are assumed to be valid during the transient state as well. a tailrace canal. 5.. the following components have to be mathematically develop the mathematical model of a hydroelectric power plant: 1. the abcissa is the unit speed. governor. *This will be discussed in detail in Chapters 10 and 11. while plotting the prototype output.3. and P in hp. and P = power output. upstream and downstream 2. The prototype efficiency is usually more than that determined from model cause of scale effects. and q are given in Table 5.2). Hn = net head. In SI units. Hn in ft. Transients in Hydroelectric Power Plants III the turbine end of the conduit system are de R6s6rvoir P6nslock Transmission lin' 5. rotational speed of the unit. is taken into account by stepping up the model efficiency. a river. q. Flow through a turbine depends upon various parameters: for example. I no data have been reported in the literature for the turbine characteristics during transientstate conditions.1. and the windage and friction losses should be known at wicketgate openings below the speednoload gate (SNL).110 Applied Hydraulic Transients Hydraulic Boundary conditions to analyze veloped in the following section. Therefore. of which the Moody tests bethis fact empirical formula? of these components are presented in the following AND DOWNSTREAM CONDUITS As shown in Fig. To do this. and the unit power. Q = turbine discharge. turbine and generator 3. then the draft tube as well as the downstream conduit system may be neglected to simplify the analysis. and the ordinates are the unit flow.4 SIMULATION OF TURBINE Turbin.N in rpm. N = rotational speed. or a short pressurized conduit. Depending upon the conduit lengths. the flow is a function of the head and the nozzle opening only. D = diameter of the runner. Thus. Q in ft3/sec. N in rpm. 1jJ.* The method of characteristics and the boundary conditions presented in Chapter 3 are used to simulate the upstream and downstream conduits. 5. As shown by Perkin et al. In this figure.Hn and D are in m. which may consist of either a freesurface flow or a pressurized tunnel. while the flow through a Kaplan turbine depends upon these variables as well as the runnerblade angle. In an impulse turbine. obtained from the model tests.3 UPSTREAM water conduits simulated to To the author's knowledge. In English units. flow should be known when the turbine rotational speed is zero. surge tanks may be provided to improve the governing characteristics or to reduce the maximum waterhammer pressures.P in kW. If the downstream conduit system is comprised of a freesurface flow tunnel. Schematic diagram of a hydroelectric power plant. an open channel. TailraC4 The relationship between the net head and discharge has to be specified to simulate a turbine in a hydraulic transient model. Various formulas have been proposed for this purpose. Therefore. to cover this range. The data for the turbine flow and power output. D is expressed in in. Curves representing the relationship between these parameters are called turbine characteristics. and the outflow from the turbine is discharged through the draft tube to the downstream water passages. and Q in m3 Is. water is carried from the upstream reservoir or canal to the turbine scroll case through a tunnel and/or a penstock. are presented in a graphical form known as hill charts (Fig. the characteristic curves are extrapolated. p. Details of the simulation sections. except for Krivehenko et aI. . Figure 5.1. SNL gate is the lowest gate opening at which turbine rotates at synchronous speed with zero output. Typical characteristics for a Francis turbine are shown in Fig. and wicketgate opening.. 5. appears to be the most suitable. Definitions of ¢.. p. and. 5. however.
5 0.50 .2 . the unit rotates at the synchronous speed. unit speed cannot 'exceed the runaway speed for a particular net head and gate opening.80 o Figure 5. Therefore.3. ~ :::) (a) Unit speed 0. " .2..06 .70 .3 0. During steadystate model tests. dunng the transient state. 0. S Runaway.OT .11 ..08 ..112 Applied Hydraulic Transients Hydraulic Transients in Hydroelectric Power Plants 113 . 5.05 ~' ~ ~ .'" '> " . At SNL gate. the curves are extended for ¢ values higher .10 Cl. '" . model data are ~ot obtained for ¢ higher than ¢ at runaway conditions (¢run)' However.09 . Figure 5./4 0. Therefore.3b) represents the conditions at SNL gate.0245 . the turbine output is equal to the turbogenerator windage and friction losses at the synchronous speed... It is clear from this figure that the value of SNL gate varies with the net head.65 . the prototype speed may exceed the runaway speed for a short duration.55 . To account for this. Characteristics of a Francis turbine (in f'tlbsec units).0/ ./3 . To keep the unit running at the synchronous speed when the wicket gates are open less than the SNL gate opening. 0../ .4 CT .. power has to be supplied to the unit from an outside source because the windage and friction losses are greater than the turbine output.. and the net turbine output is zero. if the wicket gates are steadily open at the SNL gate opening. This is called motoring of the unit..12 ~ .75 .60 . The abcissa axis on the unit power curves (Fig. Typical hill chart for a Francis turbine (in ItIbsec units).
namely. Hp=H +Ht Oalum lin.Jf1.1 (SI units) into Eq.Jf1.1 during computation of the net head. To determine the range of rP for which turbine characteristics have to be used during the time step.2) in which ao and a.<I. = = The values of four variables. if the velocity head is not small as compared to Hn. The equation of straight line EFmay be written as (5. are determined from the known coordinates of E and F. For a Pelton turbine. Hydraulic l 1 I_~__ . and the values at F are interpolated from the known values at points C and D. 5. Notation for boundary conditions for a Francis turbine. and A = crosssectional area of the pressure conduit at the turbine inlet. Substituting for q and rP from Table 5. P P (D/12)2 H~/2 P D2H~/2 X a. Q D2. Hp.nsloclr Q (D/12)2. SI Units DN 84..Jf1.. in Hydroelectric ! Power Plants liS Table 5.4. This is a valid assumption since the exitvelocity head is usually negligible. however. Qp instantaneous flow at entrance to the scroll case.ad tosses in p.Hn is also extrapolated.1) in which Hp = instantaneous piezometric head at the scrollcase entrance. English Units DN 1838.1. 5. The boundary conditions for a Francis turbine" are derived below.Jf1. Note that the velocity head at the drafttube exit has been neglected in Eq. it should be included in the analysis.3) .2 and simplifying n 31 '1 2gA2 Q~ (5. and the unit discharge and unit power at intermediate gate opening and rP values are determined by parabolic interpolation.. However. Hp. q En. Let the values of Tp. 5. than rPrun assuming that they follow the same trend as that at rP values less than rPrun' A grid of points on the characteristic curves for various gate openings are stored in the computer. extrapolation.: ~~~:~J. Let the wicketgate opening at the end of the time step be Tp. Qp. and HI! estimated by extrapolation be Te. Because the transientstate turbine speed and gate opening vary gradually. Nt!.45 . Hn = instantaneous net head. and Hne be rPe' The characteristics for Te for rP between rPl and rP2 may be approximated by the straight line EF as shown in Fig. 5. For a Kaplan turbine. these can be estimated as a first approximation by parabolic (5. and Np are unknown at the end of the time step under consideration and may be determined by the following iterative procedure. Htail tailwater level above datum. Qp. Figure 5.rgy grad. boundary conditions for a valve developed in Chapter 3 may be used. lin' H.114 Applied Hydraulic Transients Definition of unit values. The values at E are interpolated from the known values at the grid points A and B. and Hn are the values of these variables at the end of the time step under consideration. N«.:Transients :~.~: .5. :~~: . lin' Hydraulic grad. Tp. and Hne and the value of rP for the estimated values of N. Referring to Fig.4. the variation of the turbine characteristics with the runnerblade angle would have to be considered...
if Ptur and Pgen are in horsepower. then Eq.3.6) in which a4 f Simplifying. = 2a3Ca 2 a2  and 03 = ND30Ij264.5) and in which Pgen = generator load.6 yields . t P( Ptur ~ p) dt = 1. 5. 5.Tgen.672.1 0) Note that the positive sign with the radical term is neglected. and assuming that the load is only resistive. 5.14)* tl lIgen Ca Ca = 2gA 2 a~ (5.5. 3.4)* in which Ttur = instantaneous turbine torque.CaHtail) CaQ]. 5. 1.12 may be written as Pgen P tur _ _ Combining Eqs. 5.11)* c/> I Figure 5_5. (5. w = rotational speed of the turbogenerator. in rpm. Ttur  2rr Tgen . 5.619 X 106 and 0.1.13.7) (5.'" of turbine characteristics.. = power Squaring both sides of Eq.CaHn g Hn from the resulting equation (5. *If WR2 is in Ibft2. 5.(C C p a H tail  Caa~) . and replace 1.13 by 550. and Ptur in kW. (5. WR2 = total moment of inertia of the turbine and generator.15) as *In English units.15 by 0.11 through 5.3 and HP from Eq. a2 = ooD2j144. and simplifying. + ? A2 .1 . Integrating both sides of Eq.548 X 102 of Eq. 5. Now Hn is determined from Eq.18 and 5.116 Applied Hydraulic Transients Hydraulic Transients in Hydroelectric Power Plants 117 • Stored grid point o Interpolated point a6 Solu tion of Eq.9) Gate opening.r (5.3096 X 106.12)* in which and (5. Qp = lIgen _ WR2 (  2rr 60 )2 NdN dt developed (5. replace WR2 of Eqs.13 by WR2jg. then divide the righthand side of Eq. elIe Interpolation cJ>2 UNIT SPEED. N = speed. 5." Turbine characteristics (5. 5. respectively.WR 2 .13)* by the turbine. 5.097 X 102 WR2 j'NP NI N dN (5. Tgen = instantaneous generator torque. in rad/s. Because of the instantaneous unbalanced torque.548 X 102 WR2 (Nj. the speed of the turbogeneratorset changes according to the equation c T = WR2 dw u dt or (5. in kg m2• If lIg = generator efficiency.dN 60 dt (5.8) ( p turl +P turP 2 _ P genl +P genP ) 211gen ~ t = 0. .Nt) . both (Cp .097 X 102 of Eq. Tu = Ttur . .14 and 0. . eliminating Eq.
which changes the position of the pilot valve. After some time. and the accelerornetric in Europe. the dashpot spring is cornpressed. However.38 in these equations by 3. This displaces the piston of the pilot valve. and hydraulic servo.s (5J 7)* in which Pgenf = final generator load. it is proportional to n. 5% "0 Droop :: CL (f) 100 oL. then replace 182. and the oil is admitted into the hydraulic servo that opens (closes) the wicket gates. Such a governor is calIed an isochronous governor and is inherently unstable. t [ Np = { N~ + 182. Various mechanical and electrical speedsensor devices" have been used. a permanent magnet alternator feeding into a frequencysensitive network. The value of temporary droop may be large.38 WR2 0. The speed droop is a governor characteristic that requires a decrease in the speed to produce an increase in the wicketgate opening (Fig. Since a large force is required to move the wicket gates. two types of speed droop are provided: permanent and temporary. Speed droop. Of the mechanical devices. The dashpot gove. the sequence of events is as follows: The speed of the unit decreases (increases) because of the load change. then the above equation may be simplified as 6. this is not permissible from the point of view of system operation. This output is usually small and is amplified by means of a pilot valve before feeding it into the servomechanism provided for opening or closing the wicket gates.r50 100 Wicket Gate Opening Figure 5.15 and 5. *If WR2 is in Ibft2. Therefore. .5 HYDRAULIC TURBINE GOVERNORS As discussed in Section 5. a large value of speed droop is required for stability. Np = N. Permanent speed droop is usually about 5 percent. a hydraulic servo is provided for this purpose. Therefore. it is proportional to dn/dt.118 Applied Hydraulic Transients Hydraulic Transients in Hydroelectric Power Plants 119 in which subscripts 1 and P indicate the values of the variables at the beginning and at the end of time step. if the transients are caused by a step change in the generator load. a permanent magnet alternator. The output of these speed sensors is the deviation from the reference speed. (2) accelerometric. However. having a ballhead.6). the PID has been recently introduced. 5. and is utilized for sharing loads on parallel units.rnors have been more commonly used in North America. sensitivity. and a speedsignal generator. in the PID. and time integral of n (see Problem 5.3). The main components of a governor are a speedsensing device and a servomechanism for opening or closing the wicket gates. 5. Solving for Np.16. it is assumed that the generator load is gradually varied. in the accelerometric governor. even though the servo and the wicket gates are now at a different position. and Ptur and Pgen are both in hp. dnldt . 6.7) herein.2 + 182. has only one equilibrium position in which the ports of the pilot valve are closed so that no oil is admitted into the servo. and the flyballs move inward (outward). .5 (Ptur1 + PturP) .6. A governor. pilot valve.5 (Pgelll + Pgenp)J 1 }o. and (3) proportionalintegralderivative (PID). the corrective action of the governor is proportional to the speed deviation. It is made temporary by means of a dashpot.6. for stability.17 is presented in Section 5. the dashpot spring returns to its original position because of oil flow through the small orifice in the dashpot.16)* In Eqs.1)gen 0. 5 (5. Following a load increase (decrease).2.5(Ptur1 + PturP){ ~ 1)gen ] WR }o.38 2 t [O. A computation procedure for using Eqs. and ruggedness. . is fixed during a transient. As a result of the wicketgate movement. 5. a centrifugal ballhead in various configurations has gained much popularity because of its simplicity. speed droop is f"ovided. Three types of governors are used for hydroelectric units: (1) dashpot. a governor is provided to keep the speed of the turbogenerator at the synchronous speed. For hydraulic turbines. In the dashpot governor. n. The electrical speed sensors include a dc generator with permanent magnetic field.10 and 5. 5. 5. We will discuss only the dashpot governor (Fig.23 X 106.
'!? s: Ta T.. I:) . 5. ..31 through 5.. NR . ~ ~ '0 ~• CI. ~ .8. n = N/NR.e. Typical values of these constants are: Ta = 0. .8...0 + 010.0) • .. then Ilref = 1.. In the transfer functions. ~ ~~ .7. In this diagram.5 .8 shows the block diagram for a dashpot governor. is used to normalize the turbine speed. I~c • • '0 The synchronous speed of the turbogeneratorset. s is the Laplace variable. 5. ~ ... Figure 5.  La 1)... If 10 = initial steadystate wicketgate opening. = co '0 . = actuator a . The following notation is used in the block diagram shown in Fig..ls 0=0)3to0. 5. i.8: time constant dash pot time constant o = temporary speed droop 0= permanent speed droop Td = distributing valve time constant kd = distributing valve gain ks = gateservomotor gain n = normalized transientstate turbine speed.5 0. and the transfer functions listed in the blocks show the relationship between the input and output of various components. ~ . c _'<t ..120 Applied Hydraulic Transients Hydraulic Transients in Hydroelectric Power Plants 121 spring Pilo/ Oosllpo! To op.. Input and output of various blocks is shown by means of arrows.05 to O.n go/. then s is equivalent 7 to the time derivative d[dt .... a .34..... Figure 5. The outputs of various governor components and their saturation limits are shown in Fig. .. '<t . Readers not interested in the derivation may proceed directly to Eqs..~ a ::. :~ . 5.. The differential equations for different components of the governor can be written using the transfer functions listed in Fig. If initial conditions are zero.. Dashpot governor (permanent speed droop not shown)..0. conventional notation of control systems" is used with different blocks representing various components of the governor.+ I..
and rearranging Eqs. 5.OS to 0.21) equations may be written (S. .122 Applied Hydraulic S Transients Hydraulic and Transients in Hydroelectric Power Plants 123 Td = O.33) (5.11. and 5.1 kd = 10 to IS ks = 0.vdl dr dt = ksvd 1 Vdmax = k T s (5. By eliminating e and ed from Eqs.S v dmin : I k s Tc (S ..22) Note that the output of various components may saturate and that these saturation limits must be taken into consideration in the analysis of large load changes. I Gate Servomotor v =e a Tas or (S .25) 0 The preceding four differential equations in four variables.29) or Vi = Va  r (S.30. 5.26) For a given set of power plant parameters.29. a cIosedfor:n solution is not possible because of nonlinearities introduced by the saturation of .k Vd dt S et = I + Trs va =0 0 ~ r ~ 1. The following (S .r) .34) (5.n .22.e ) r dt t (5.18) (S. namely. eliminating Vi from Eqs. which are defined as twice the time taken by the wicket gates to open or close between 25 and 75 percent openings.ava) dVa .28) s r. may be selected by using the procedures outlined in Section 5.24 and 5. Vd.2 etmax = 0. and t .24 and 5.. in the above inequality.28. S.31) (5. the optimum values of 0 and T.19) or dr .0 because of two feedbacks (S. we obtain Distributing Valve dt = T (nrer a dVa I .32) (5. Actuator in which To and Tc are the effective wicketgate opening and closing times.24) The gateservomotor rate limits are often applied by restricting the maximum travel of the distributing valve in the positive and negative directions.27) e = T dVa a dt Dashpot (S .2 to O.23) (5. may be integrated by any standard numerical technique. we have det = __!_ dt Tr' (oT dt dVd I = Td [kdeVd . 5.30) Permanent Drop (S .e. Therefore.20) and (S.19. va' et.
5. now being known. 5. 5.9.315.7 CAUSES OF TRANSIENTS The following turbine operations produce transientstate conditions in the water conduits of a hydroelectric power plant: Figure 5. The coefficients ao and 01 are computed. If l'Tp .4 and 5. . and the computed value of Hn. then Ne is assumed equal to Np. the grid points A through D (Fig. A computational procedure for using these equations is presented in this section. and the generator load.124 Applied Hydraulic Transients RungeKutta method" in our Hydraulic Transients in Hydroelectric Power Plants 125 various variables.1 0 is used to determined Qp. Note that this equation is valid only if the time steps are equal. 5. For the estimated values of Hne. Since there are no previous time steps at t = 0.17. Ne. Now Eq. otherwise. The flowchart of Fig. Let us assume that the transientstate conditions have been computed for (i 1) time steps. To avoid unlimited repetition of iterations in the case of a divergence in solution. 5. and the values of coefficients a2 to a6 are determined from Eqs. Flowchart for boundary conditions for a Francis turbine. 'Te. PturP.002 NR. Pgenp. time is incremented. and the above procedure is repeated.35) in which Y is the variable to be extrapolated? and the subscript indicates the time step.3Yi2 + Yi3 (5.34) are solved for 'Tp by the fourthorder RungeKutta method. and the above procedure is repeated. We have used the fourthorder analysis. 5. 5. The values of Ne. and Hne at the end of the ith time step may be estimated from the known values for the previous three time steps by parabolic extrapolation from the equation Yi = 3Yi1 . and 'Te. the governor equations (Eqs.'Te I> 0. 'Te is assumed equal to 'Tp.7 through 5. PturP. and Eq.5) are searched from the stored characteristic data. the steadystate values may be used for previous time steps for extrapolation purposes. a counter is introduced in both the iterative loops.6 COMPUTATIONAL PROCEDURE Boundary conditions for a Francis turbine and equations tor a dash pot governor were derived in the preceding sections. and the transient conditions are computed for the system.9.9 illustrates the preceding computational procedure. is then determined from the turbine characteristic data for rfJe and 'Te' The turbine output.3 to determine H fl' The value of rfJe is computed using the estimated value of N. otherwise.005. 5. 5.16 or 5. If INpNel> 0. and the turbine output. the value of Np is determined from Eq.
and the load was rejected. Gates are usualIy kept at this opening until the unit speed is about 60 percent of the rated speed: then the gates are closed to speednoload gate. The power plant consists of 10 Francis units. It is then synchronized to the system and is ready for load acceptance. and a freeflow tunnel conveys water from each manifold to the tailrace channel. which was free to turn about a vertical pivot anchored to the upper bearing of the turbine. A tensioning device held the drive wheel of the tachometer in contact with the turbogenerator shaft.1 s. The friction factors were calculated such that other minor losses were included in the friction losses. the gates are closed from one opening to another for a load reduction. and 250 MW were con ducted. and was kept at this load until pressure and flow at the turbine inlet became steady. The wicketgate opening was recorded as follows. The Westinghouse leadingedge flowmeter? was used to measure the steady as well as the transientstate flows. five units discharge into one manifold. Speed changes should therefore be taken into consideration in the transientstate computations for these turbines.10. to simulate an isolated load rejection. However. load acceptance c. Then. the flowmeter digital display along with clock time were recorded on a videotape. load acceptance b. depends upon the type of rejection. which was proportional to the turbogenera tor speed. The upstream reservoir level during the tests was at an elevation of 671. the speed of an isolated unit rises during load rejection and decreases during ~oad acceptance. the wicket gates are opened at the prescribed rate to the opening at which turbine output will be equal to the final output.49mdiameter pipe. load reduction or total load rejection 2. 115.* The length of the scroll case was taken as onehalf of the actual length to account for *See Appendix A. The speed of the turbogenerator was measured by a de tachometer generator. A unit connected to a large system runs at the synchronous speed during load acceptance or rejection because of the large inertia of the system. M. The motion of one of the servomotors of the wicketgate moving mechanism activated a precision voltage divider (potentiometer). Shrum Generating Station. was recorded on a Sanborn recorder.126 Applied Hydraulic Transients Hydraulic Transients in Hydroelectric Power Plants 127 I.4 of the G. unit startup b.6 are for the isolated units only. The unit was loaded to the amount to be rejected. Similarly. In the tailrace. however. The conduit between the trashrack and the downstream end of the transition was replaced by an equivalent 5. the turbine speed has a considerable influence on the transients. Each unit has its individual power conduit (penstock and power intake). The flowmeter display exhibited average flow every 2. A rubberfaced drive wheel was fastened to the shaft of the tachometer. on which tests were conducted.202. when appropriately conditioned through its straingauge amplifier. The boundary conditions and the computation procedure described in Sections 5.2.4. Data for Unit No. As observing and recording the flowmeter readings at this rate was difficult. These conditions can be used for . and of the downstream water passages in Fig. Canada. The output of the transducer. Shrum Generating Station is located on the Peace River in British Columbia.0 m. A schematic of the upstream water passages is shown in Fig. load reduction or total load rejection. 118. The change in voltage was then recorded on an oscillograph.2 m.4 through 5.8 VERIFICATION OF MATHEMATICAL MODEL Prototype Tests To obtain data for verifying the above mathematical model.10. the tape was replayed at a slower speed in order to note the readings. The voltage output of the tachometer. was recorded on a chart recorder. Prototype Data G. are given in Table 5. M.units connected to a large system by keeping the speed constant and by bypassing the loop for computing the speed changes. 5. Wicketgate closure following total load rejection. For Kaplan and Francis turbines. While starting a unit. the speednoload solenoid was blocked. After each test. . 12. 5. The transientstate pressures were measured by a straingauge pressure cell attached to the turbineinlet piezometer manifold. and the unit is allowed to run at the synchronous speed for a short period of time. Unit synchronized to a large system a. Locations of the flow transducers are shown in Fig. Isolated unit a.20. loadrejection tests were conducted on Unit No. The tachometer was mounted on a horizontal arm. and the downstream manifold level was at an elevation of 503. owned and operated by British Columbia Hydro and Power Authority. Five tests involving load rejections of 50. 5. For load acceptance. the wicket gates are opened to the breakaway gate opening to give the unit a "kick" to overcome static friction. which was calibrated against 0 and 100 percent gate openings.
.. .0s 0. Shrum Generating Station...e <::> . ..9 Friction Factor 0.. . 0 Pipe No.49 4.5.~ ~ . Static head and an estimated initial .... etmax Selfregulation constant. . Comparison of Computed and Measured Results A computer program based on the preceding mathematical model was developed. i. Waterhammer wave velocities were computed using the equations presented in Section 2. ~ ..27 Gg m2 (i.27 4.:i :i ..009 .15 Conduits Length (m) 207 78 36.4 0. The downstream manifold. 41 o e 2 en . and the turbogenerator and governor were simulated using the equations derived in Sections 5. I. Pipe No.6. ...~ " ~a ll: ~~ ~ ~ t:. The draft tube was not included in the analysis because of its short length . . a time interval of 0..!. was assumed as a constantlevel reservoir.. Hydraulic Transients in Hydroelectric Power Plants Data for Unit 4..49 5....86 m X 106 kg m2) • 0 .. . To satisfy the Courant's condition for stability of the finitedifference scheme. ~ ~ '"' <l iE '" <::S Q:: 0 II') "" ~ .e..4 and 5...~ i: ~ S .~ :i ED 0 231 MW 152.128 Applied Hydraulic Transients Table 5. 1 2 3 Diameter (m) 5.:: 129 .e.c c. ~ t. li Permanent droop. .. !).016 0. The flow in the conduit was analyzed using the method of characteristics of Chapter 3.. ~ .3 was divided into two reaches. T.. G.5 Wave Velocity (m/s) 1244 1290 1300 '" :§ : 1U6"·~ ·r ~ :.. " '" '" t.. This simplification should not introduce large errors. en '0/0 ~ Dashpot time constant. .. 0 "0 "0 41 41 ] . Temporary droop.4 m 150 rpm 164m3/s 9... Turbine and Generator Rated turbine output Rated head Synchronous speed Flow at rated head for rated output WR2 of turbine and generator Runner diameter Ii. 9.010 0. M.05 0. The wave velocities in the upper pipes were slightly adjusted so that there was no interpolation error. :: ~ ~ . being a freesurface area.. " ~ I.§ reduction of flow along its length.25 0..t ~ Sx]«. a = c Governor Settings 8. E E 0 . 0 Dashpot saturation limit.014 s was used.2. '" .
2.. The computed pressures show some oscillations that were not recorded during field measurements. . the computed results show a faster speed reduction than that shown by the measured results.. Normal All operations that are likely to occur several times during the life of the penstock are termed normal. 4 Time after 6 =:::_ ~ _  Measured Computed ~ 8 10 load rejecticM. If the power output was different from the actual value. As the unit was assumed to be isolated from the system following load rejection. 3.!> . " "\ \7 10 . the computed and measured pressures agree closely. appurtenances or devicessuch as surge tanks. the initial steadysta te gate opening.9 DESIGN CRITERIA FOR PENSTOCKS As discussed in Section 4. the initial gate opening was slightly changed. 60 ~ 40 r.. t = O. ~\ "" . Openi~g of the wicket gates from the speednoload to full opening in effective gateopening time.e. .. seconas I 210 '5 20 Measured _ Computed ~ Il: 195 /j / /" ~ ~ . <t 180 175 170 165 fl r 0 'i # Y II \\~ \ . Computed turbine speed. i. emergency. however. Load was rejected at time...J 8 load rejection. The cause of this difference has not been explained.11 and 5.12.7.130 Applied Hydraulic Transients Hydraulic Transients in Hydroelectric Power Plants 131 steadystate gate opening were the input data to the program. or catastrophic . secoflds 12 14 Figure 5.  M~asured . after every 0. As can be seen. *Effective wicketgate opening and closing times are defined as twice the time taken bv the wicket gates to open or close between 25 and 75 percent openings. nor do they drain. Fullload rejection and closure of the wicket gates in effective gateclosing time * with the sta tic head on the turbine up to its maximum value. was determined. gate opening. The computed and measured maximum unit speed agree closely. which gave the required turbine output./ 12 " 14 seconas 190 . This difference may be due both to an error in the estimation of the windage and friction losses and due to lack of data for the turbine characteristics at small wicketgate openings. and the above procedure was repeated. . During these operations. 5. flow.10. 4 Time after 6 8 10 12 14 0 load rejectiofl. . Based upon the frequency of occurrence.. By using this trialanderror procedure. .. The following are considered to be normal operations: I. 150MW load rejection: comparison of computed and measured results. Corresponding flow and turbine output were computed from the turbine characteristics...11 A discussion of the operating conditions included in each of these categories and recommended factors of safety"? follows. 2 ~ ~.e ~ 160 150 0 / /1 2 ~ . and cushioning stroke devicesprovided for reducing excessive pressurerise or pressuredrop function properly for which they are designed. various operating conditions may be classified as normal. pressureregulating valves. unless an overflow weir is provided. with the static head on the turbine as low as its minimum value. the factor of safety to be used during design depends upon the risks involved and the probability of occurrence of a particular operation during the life of the project. 5..11.Computed ~ ~. The surge tanks do not overflow. \ \ '\ 2 4 Time after 6 I \~'\ \J ''.5 s of prototype time. and pressure at the downstream end of each pipe were printed after every 35 time intervals. it was allowed to overspeed.!> " . . 190 ~ 185 . Computed and measured results are plotted in Figs.. and the wicket gates were closed under governor control. It should be noted that this deviation starts when the wicketgate opening is small....
 ~ ~ ITO V / 7 h/ /' i= _ _ r.!: 80 ~ tJ til c:: 0:. Turbine speed 2S0·MWload rejection: comparison of computed and measured results.......!. !'. is recommended for pressures produced by emergency operations.Compulfld f.'8 load rejection.' 190  ~ . '\ \\ \ \ \1/ 8 <... The cushioning stroke device is inoperative on one unit... Although the increase in the generator cost due to increasing its inertia may not be large.12. " 1".. 2... only the latter is increased. it is necessary than an adequate amount of generator and turbine (unit) inertia be provided.' tJ 200 19~ 190 18~ 180 c:: ~ ~ tl  ~ ..'8 load reJection. Mflosurfld ./ I' 7// 1\\  Meosurfld Computed _ Catastrophic Catastrophic conditions are those in which various control devices malfunction in the most unfavorable manner. if the pressure regulating valve malfunctions following a load rejection and the wicket gates do not close at a slow rate.. Hydraulic Transients in Hydroelectric Power Plants 133 * 100 .. 12 I4 sflconds 16 /.132 Applied Hydraulic Transients ~..... a factor of safety of slightly more than one. t/'J \. A llowable frequency fluctuation. A factor of safety of two.. other associated costs.1 0 GENERA TOR INERTIA For stable governing of a hydroelectric power plant and to keep the speed rise of the unit within permissible limits following a load rejection. . ". The following factors are considered in selecting the generat or inertia: 1.. The pressureregulating valve is inoperative on one unit...'8 ~OO The emergency conditions are those in which one of the pressurecontrol equipment malfunctions. based upon the ultimate bursting or collapsing strength.... 20 2 4 ~ ""~ 6 Timfl of/fir (0) 8 10 load rfljflction....1 \x. The allowable frequency fluctuation depends upon the type of load. (b) 210 Penstock pressure ". IT~ ITO 16~ 160 0 ! /1 if 1/ /1 . then this operation is considered catastrophic_II Because of very low probability of occurrence.. 10 . Therefore. based on the ultimate bursting and collapsing strength. Increasing the generator inertia increases the cost of the project....tI _ ~ Measured Computed I 2 4 6 Time oflflr 8 10 12 14 16 /. ~ . These conditions include: 1. is suggested.. then the wicketgate closing mechanism is designed such that the wicket gates will close at a very slow rate in case the pressureregulating valve is inoperative during a load rejection... secona« (c) Figure 5. such as increasing the crane capacity or increasing the powerhouse dimensions. a frequency deviation of 0.. . j . based on the ultimate collapsing or bursting strength.... For example. if necessary. . _ :} 60 40 The penstock and scroll case are designed to withstand the maximum and minimum pressures caused by the preceding operations with a minimum factor of safety of four. 5. r12 seconas 14 f_ 2 4 6 Timfl of/fir 16 /."\ . therefore. Turbine Qote openinQ 210 20~ l:. For example... if a pressureregulating valve is provided... Emergency (Co~1 A5t'1[ ... a~e usually high. the generator inertia is kept as small as possible while still maintaining acceptable governing characteristics. Turbine inertia is small compared to the generator inertia.. However.
the generator inertia may be decreased. the stability of the system can be improved.6 X 106 hp). are turbine flow and net head at rated conditions. For good regulation. . Tg. however.970 (:. L and A are the length and crosssectional area of the water passages. and layout of the water passages of the power plant.. 2. A unit should be designed to be stable in isolated operation if it is supplying 40 percent or more of the system load. Therefore.4 through 5. they cannot be arbitrarily decreased since they are set so that the waterhammer pressure is within the design limits (Section 5. Gordon's stability curves should be used because they take into account most of the factors upon which the generator inertia depends. = synchronous speed. Normal generator the flow from zero to rated velocity. Type of load.38) in which g = acceleration due to gravity. penstock. in rev/min. the United States Bureau of Reclamation recomrnendsP that the ratio of the mechanical starting time. To use these curves. By decreasing the governor opening and closing times. Experience curves proposed by the Tennesee Valley Authority relate Tm and show stability limits for various ratios of the unit size to that of the system. Paynter!" presented a stability limit curve and suggested optimum values of the governor settings by solving the problem on an analog computer. and the required generator inertia is then determined. N.000. such as electric trams and mining shovels. kua = the generator rated output.17 derived a similar stability curve theoretically. Twand 5.5 s) to the effective gateopening time. the mathematical model developed in Sections 5.134 Applied Hydraulic Transients Hydraulic for the rated head to accelerate sions for Tm and T ware Transients in Hydroelectric Power Plants 135 Expres percent is not permissible for paper mills. both Paynter and Hovey neglected the permanent speed droop of the governor and the selfregulation of the turbine and of the load. However. Gordon!" has taken into consideration the effect of the governor times while plotting his curves (see Fig. replace 15. During the final design stages. One of the major factors in the selection of the inertia is the size. the wicketgate opening time. and its length is taken as onehalf to account for the reduction of flow along its length. However. which are based on experience with 40 Kaplan. and governor.37)* Tw =~ (5. or an accelerometric governor is provided to control the speed oscillations of a turbogenerator of a hydroelectric power plant. Hoveyl6.)i is not zero 'ijJeeadroopa *If WR2 is in Ib ft2 and the output is in hp. and T w is the time *In English units. No analytical method is available for determining the generator inertia required for a given set of plant parameters. Francis.. permanent . Therefore.:. a dashpot. In the author's opinion. length.13). As discussed previously. Tm' and the waterstarting time. and so that the water column does not separate at high points of the penstock or in the draft tube. However.~) 1. The normal or standard generator inertia depends upon the unit rating'? and is given by the equation. the size of the water passages is first selected based on the costbenefit ratio of reducing the head losses. Periodically changing loads. contribute to system instability. the former is usually more costly. T w. is computed by adding the time of the cushioning stroke (about 1. Tm is the time in seconds for the ra~ torque to accelerate the rotating masses from zero to rated speed. and ~ LIA is computed from the upstream intake to the downstream end of the draft tube. The crosssectional area of the scroll case at its upstream end is used for computing ~LIA. and WR 2 = moment of inertia. be greater than 8. In most cases.9). 4. Size of the system. Units with Tm /T w of 5 or less may be integrated into a system but it may be necessary to compensate this deficiency on the other units of the system. a PID.25 (5.6 should be used to confirm the results of the preliminary analysis. more inertia should be provided if such loads are present in the system. T m = WR2 X N2 r 90. 3. and propeller turbine installations.4 X 106 MW Qr L gHr A (5. To. Qr and H. The speed oscillations are stable or unstable depending upon the values of the parameters of the hydrounit. while a deviation as large as 5 percent may be allowed for mining equipment. 2. 5. Water passages. a number of empirical formulas and experience curvesl214 have been proposed. the value of WR2 may be increased or decreased.36)* in which N. then T m = WR2 X Nt /(1. = the synchronous speed. The overall stability of the system is increased if the majority of the units in the system are stable in isolated operation. By increasing the size of the water passages.970 by 379. Governor times. in kg m2 • Depending upon the factors just outlined. or if there are possibilities of the unit becoming isolated because of failure of the transmission line. Therefore. 5.11 GOVERNING General Remarks STABILITY WR 2 = ) 5 .
.136 Applied Hydraulic Transients ... if a and a are allowed for..0> ::z ~t~ !:~~2 i'. . or hysteresis . the following differential equations! 5!8 can be written for the hydroelectric power plant shown in Fig.. By making these assumptions.: II ._.18 mayor may not be zero depending upon the type of load.... head...5h  em  ·f:. Thus...) The stability criteria are formulated herein!9 by taking into consideration the permanent speed droop and selfregulation... 5. . <: .... " ·s .. 0 I 0 oJ .I . ~~t ~! ii " . . The load selfregulation coefficient 0<[.!i: Hydraulic Transients in Hydroelectric Power Plants Table 5. ~~~e ~::~= rr r..39) .. see Table S. 2. z"'II:~ io i:: 0 "i! 0 0 ~ i "0" ~~ ~'g~ ~ .. Tr... '"  N z . ~ l} . z 0" ....~~ .~ Turbine In general High specific speed Load Grid loading: Motors only (constant torque) Ohmic resistance only with voltage regulation Ohmic resistance without voltage regulation 8Taken from Ref..6 .." 1.....'" . and gate opening are small.. the stability is considerably increased. O<tup is defined as the slope of the graph relating the per unit deviation of the turbine torque from the rated value. 0 '" . The walls of the penstock. OL/ OLturb OL = OL/  OLturb " ~ ~ " ti'· !ii b~ ..~ s I o +1 0. and the water in the penstock and scroll case are rigid....~ ~ " ~ & . o .r.. waterhammer pressure caused by changes in the gate opening can be computed by using the rigid watercolumn theory. thus. o ::= " ~ C = :E "' ..... ~I 0 .. while theselfregulatlon coefficient... .. Values of selfregulation coefficient..3... The selfregulation coefficient is defined as the algebraic difference between the load selfregulation coefficient and the turbine selfregulation coefficient.L N . are reduced ..1 ._lt . . i~~ ~ .. .. 3... to the per unit deviation of the turbine speed from the rated value... T m dn . 0 o "'The turbine selfregulation coefficient.3.. j:. Differential Equations of the System In formulating the differential equations. ... The changes in the turbine speed..... .... =!§ z ex "0 ~ . .. z .. backlash. is defined as the slope of the graph relating the per unit deviation of the torque of the electrical load from the rated value.:: ~~z .* (For values of a... to. the following assumptions are made: I. and the dashpot time constant.. I:i ..0 2 to 5 I to 4 e ........ and the optimum values of the temporary speed droop. It is found that.. 1.. a 137 .~ ~ i:j! ~:... ." i.. ~ 0 0 Z . . ~ ~ . to the per unit deviation of the frequency of the electrical load from the rated value...r. Machine acceleration r ~~~~ <. a..... about I up to 0. . The governor has no dead band.."' "i! 0 0 ! . 18. .=g dt + 1.m (5.. oJ . I· I' . '" sc "'e ~ .. or. 0 0" . . r 0 I 0 oJ z e .. A single hydrounit supplies power to an isolated load . 4... nonlinear relationships can be assumed linear. 5..: ~~: . Q....
g = relative gateopening change (G .138 Applied Hydraulic Transients Criteria for Stability Hydraulic Transients in Hydroelectric Power Plants 139 2. .5 T ws(Trs + l)n [(a+6)Trs+a] (0. Then Eq.52.52) (Trs + I)n T pn m (a+6)Trs+a By simplifying + 1.Tw + 0.51 are always satisfied. 5. !lM = stepload torque cha~ge (negat~ve for load rejection).Sm (5.2o the oscillations differential equation (Eq. Water acceleration dh dg 0.5 Tws+ I) . the following nondimensional parameters are introduced: (Trs+I)n g = (a + 6)Trs + a It follows from Eq. Let s = dfdt The subscript 0 refers to the initial steadystate as values. 5. S2 by d2/dt2 . h = relative pressurehead rise (H . M = initial steadystate load torque. limit of stability: into inequalities 5.4 I may be written [(a + 6)Trs or + a]g = (T. .« + I)n h = ~~:=:[(a + 6)Trs + a] (0.. Tw' Tm' and Tr. d3n + [0. = = =: 3. 5.O'n .5 T w r.43 and 5. Tr{a + 6)] (1 + aO').5O'TwTr(a+6)] > [0. and s by dldt .4 I) (5. (I + O'a) > 0 >a (5. .40) According thirdorder to the RouthHurwitz criteria.52 and we obtain the following equations for the 5.47) (5.5 O'aTw + (a+ 8)O'TrJ 4.Tw + 0. To plot the stabilitylimit curves.45 into Eq.48 and 5. [aTm + T. we have to consider the expressions given by inequalities 5.5 Tws+ or I)h = Twsg (5.46) and replacing S3 by d3/dt3 .46 takes the form 05 T w T m T r (a+li)3 dt .5aTwTm +(a+6)TmTr ret.5 aTwTm (a+6)TmTrTwTr+0.48) n (5.Ho)/Ho.50) (5. andN = transientstate instantaneous speed of t e tur me.5 aO'Tw + (a + 6)O'TrJ [0. and 5. relative loadtorque change = flM/Mo.« + 1)11 5. a.53) + (I + aO')n = aflm (5.Go)fGo.45) (5.5 T w dt = T w dt + h 3. 0'.43) (0. 5.40 that (5.44) Tw si'I'.50.47) are stable if represented by the dg (a+6)Tr+ag=Trdt d dn t (5. 5.54) . (5.42) The inequalities 5.39 yields (5. 6. namely. [aTm + T. To reduce the number of parameters and to present the criteria in a nondimensional form. 5. Governor response (5.51) + (5.5 T wS + I) Substitution of Eqs.49) in which n = relative speed deviation = (fV . G = transientstate instan.49. Eq. There are six parameters in these expressions.50.49.52) in rpm. 5.a h bi taneous gate opening. By substituting the above parameters simplifying the resulting expressions. (5. and 5.No)/No.
A3 = 0. for AI = 0../Jle illustrates that the results obtained by neglecting a and 2. 5..035.55 and 1.05 .4 1.1 1.2 .035 X 9.8 ).4 0.25 s. A.28X9.1. 1. This is shown as a dotted curve in Fig. ex may be taken equal to 1 (see Table 5..24 0. Example For 5. T w = 1.4 .j + A..4~r.3 = 0 and 1. Transients in Hydroelectric Power Plants 141 AfO'4 + O.4=::: r.0.2 O.. Based on these equations. The following example ex are conservative.25A~ + (1. are stable. oscillations are stable . 1.05 s.49.3 + A.6 \ l/jTm  Tw 1.0.140 Applied Hydraulic Transients Hydraulic 1.255.05 2.2 00 04 ).14. the Kelsey hydroelectric plant.4 = 0.~ + A2A4 + 2A..22 a = 0.4 0. according to his criteria.. Hovey's stability curve is obtained.4 (c) 1. which lie in the region enclosed by the stability limit curve and the positive coordinate axes..5A. oscillations are unstable (Hovey's 2·1.4 I were solved on a digital computer. r.3A.0 I.3 + 1.0. =0.4 1. «r.3 = 0.. For 1.. Stability limit curves.OandA.3=0.3 =0. o 0.4 0..4 SIa. Speed oscillations corresponding to those values of Al and A2..3A. The Kelsey plant supplies power to an isolated load consisting of furnaces.2 ).21+1++1 I.55) 0. The plotter output obtained from the computer is presented 1... 0...8 .4 1...24 1 X 1.2A. the stability limit curves for different values of AI.A2A3A4 + A3A~ + A2A3A. Eqs. «r.0 0. 5.5A. Hovey reported'" that.55 represent the stability criteria. it follows that. A3 = 0.3== 0. i \ L < 0. Stability calculations can be done as follows: 1\ 1.SA3 .4= 0.2 1.255.2A~) A. 1.... 0 I+'ftI. r.51.24 sand Tm = 9. the speed oscillations caused by a step load change are unstable for /j = 0.' 7 Thus. The following analysis shows that the oscillations are stable for these values of /j and T.and 1.25i\2A~A4) (5.31.6 < t 0.55 r. 049 : 0. 3.39 through 5. 5.137 0.2 ===0.2A4 .2A~A~ + A I (I 1.4 are plotted in Fig.~ + 0. oscillations are stable criteria) To check the validity of these results.0.5 AD = 0 Equations 5.14.~A.2 A.3 + i\~A4 + 0.I + A.1 and A4 = 0.8 " 0. oscillations are stable 4.25 =0.05 Figure 5.28 and Tr = 2.0. and compressors.4 = 0.2.. 5. if the permanent speed droop and the selfregulation are taken into consideration.25A.3)..2 = 0. 1.~ + 0.5A~A4 .4 = 0.2 0. 1.2A. blowers. As reported by Hovey in another paper .53 through 5.0 and 1.24 From Fig.24 9.14.255..3.14.5A2 + 0.5A~A~ .1 and 1.
.000 .02 . and 5.24 . .62) Transient Speed Curve The differential equation.02 0 20.02 0 . " " ~ "" ~ ~ .225 Hydraulic Transients in Hydroelectric Power Plants 143 .225 By substituting Eq. we obtain dgl (a + 0) T.n=np+nc)·ForEq.l:! By substituting these values into Eq.0 (C) 40.5.47. 5..: ' .otsm . Solution of the Differential Equation.06 .N =No. . I t=O (5.490 .34 .. ~ " ~ "" ~ .!? . A3.59) " .490 .0 Time (Sec) (d) Unstable and stable speed oscillations.137 .. np (Le. 5. yields Differentiation of Eq. we obtain dnl Tm dt or (5.15.. G = Go.551 .04 LI L2 L3 L4 = = = = . dhl =_2_ dnl dt t=O (a + 0) dt t=O Substitution of Eqs.57) Note that for load. L2.61 into Eq.04 (0) .~ " .60..08 .57.0 40.61) Figure 5. ... . Eq. hlt=o =0. " ...) . L3.04 .56 into Eq.000 LI L2 L3 L4 = = = = .0 0 20. 11m is negative... 5..58) .. .39.5 dt dn 0:dt (5. ~ "" ~ .551 L3=.. 5.0 Tim. nco and the particular integral. resulting equation yields d2n dt2 (5. and L4 denote AI> A2. nlt=o =0.490 L2 = .551 . and ~.0 60.04 0 ..ao: . and H = Ho' Therefore..47.60) lo.~ " . glt=o =0.00: (11m) a+" U Tm2 in Fig. 5. In this figure. 5.137 L4 = .06 .000 r 60.47 The general solution of the thirdorder differential equation. a = . 5.02 0 = 11m t=O II:: . 5.. 5. may be solved as follows to determine an expression for n: Initial Conditions At time t = O. 5.40 and 5.60 that (5.551 L3= . . " .~ LI = . Eq.14 .Tr.04 .000 L4 = .0 60..56) (5.06 .5.0 (5..~ ~ " ~ .12 .0 40.0 Time (S..490 L2 = .04 .0 Time (Sec) 6QO ( b) T m =+ 2 d2n dt dg dt dh 1.dnl dt t=O (5. dt t=O which upon simplification becomes dgl __ _1 dnl dt t=O (a + 0) dt t=O It follows from Eqs. ss«) . respectively.47.c) . is equal to the sum of the complementary function. " "" ~ " . " ..off condition.63) .142 Applied Hydraulic Transients LI = .15.08 .58 and simplification of the = 2 . n =p (1 + ao:) .08 0 20. ' I t. Ll.~ " .0 40.39 ./6 0 20. Eq.41. 5.
and using the initial condi 1.3 + [0. and using the initial condition .24 = 0.137 r.25.255.3 and 1. 1.137 and A4 =: .4 = 0. 5.3 = = «r w = 1. 5. and E are arbitrary constants. Then.16.2.2 are considered optimum values. . The following example illustrates the procedure to select the optimum values from this figure. B = [. nc =Ae'>r + De({3+iy)t + Ee«(3iy)t where A.05 aT m 1.5 TwT mTr(o + a) p. 5. B.. compute the roots of the char Hydraulic Transients in Hydroelectric Power Plants Solution of Eqs..5 T wT m a + (0 + a) TrT m .144 Applied Hydraulic Transients function. The optimum values of 0 and T.Am _ (a' _ /3)A _ a/3Am] _!_ r.4 : A + C= By differentiating Eq. substituting given by Eq..SexTwTrCo a 2 2 (a + o)T~ . may be determined as follows: a.66) where A.67. D.16: From this figure.2 from Fig.2 n Hence.Tw + 0. Determine the optimum values of AI and 1.64 be ex'.69) t = 0.65 can also be written as (5.4 = = r.24 0..67 twice. and C yields 145 To determine the complementary acteristic equation. 5. I t=O oSm =A + C=: 0 (I + aex) (5.035 X 9. 0. The optimum values of AI and A2 for different values of A3 and A4 are presented in Fig.0 X 1. we obtain aAm (l + co) (5.(I + aa) (1' + /3 ) (a'/3)2+'Y2 (Am) (5. 1+ oa Optimum Values of the Governor Parameters (5.05 s (computed from the known value of WR2 and ratings of the turbine and generator) o= 1. This procedure is repeated for A3 = 0.430 and 1. B.24 s (computed from the dimensions and geometry of the penstock) Tm == 9.aex . 5.0 and 0.5 Twaex (5. 9.035.~ (5. a+5 T. we obtain a'2 A + 2{tyB + (/32 _ l' (5..255.oex (Am). and 1.71) b.68) Determine the optimum values of 0 and T.ex] J. /3 ± i'Y... substituting t tion given by Eq. but slightly underdamped response..69 to 5.3 = 0. 2) C= 2 . which give the shortest settling time. Eqs..1 + (l + co) = 0 Let the roots of Eq. 5.2 = 0.TwTr + a)] J. 1 +aa l' C=A.39 through 5 AI are solved for various values of A I and 1. Example 5.65) Equation 5.12 + [Tma + T. and C are arbitrary constants.0 (determined from Table 5..67) The values of the arbitrary coefficients are determined from the initial conditions For a specific value of A3 and a specific value of A4.72) + (0 + a) T.05 0 = . Tm By differentiating Eq.(a+o)ex + O.62. ).3 for the type of load) a= 0.0 to 004. Those values of A 1 and 1.71 for A. • a Therefore. The following are the values of different parameters: T w = 1. 0.. the complete solution is n =AeO! t + e"t(B sin 'Yt + C cos 'Yt) .(I + aa) oSm (5.27 for 1. AI = 0.64) A= r::: + otsm 2/3 2 . I. 5. Am ex A +'YB+/3C==.70) 0. for Kelsey hydroelectric power plant. 5.57. Compute 1.
.= 7.137 L4=.255 ::~ it I :::I .32 '~ . It is clear that.0 40.0 40.24 Tr = .137 L4=.146 Applied Hydraulic Transients c~ I Hydraulic Transients in Hydroelectric Power Plants 147 '~ ~ .3::0 ~ 8=~= 0. 3./6 ~08 ~ II) Q:: 0 20. 5. governing stability studies carried out for the Kootenay Canal Development are presented in this section.24 _ 0. Optimum governor settings.j 0 20.96 s /I .17 0. H in ft.0 e ~ II) .12 CASE STUDY For illustration purposes.4.27 . generator.050. Speed deviation for various governor settings. Optimum values suggested by Hovey: 2Tw 8 == r. and In the English units..08 " ~ ~ ~ .3 04 For cases 1 through 3.0 60. units. Compute 8 and T. 1.JP/HI.255 038L_~~_LJ ).1 0 0. c. There are four units in the plant with each unit having its own power intake and penstock.25  ~ .= _. in SI in rpm. 2 X 1.24 ~24 ::'_' .0 40. Data for the turbine.3=0. 5.4 T m 1.6 s 2. if a and 0: are taken into consideration.400 L2=.3 TIME (seconds) (b) TIME (seconds) (c) Figure 5.17 0.0 60.24_ r A2 0.16. H in m .4 0.500 L2=. P is in kW. Kootenay Canal Development is a 500MW hydroelectric power plant owned by British Columbia Hydro and Power Authority.0.4 X 9. 1. 209 (SI units) *Specific speed = n.3 s 0.24 9.24 =0.17.0 0..1 are presented in Fig. n .319 T = Tw = 1.24 = 4.270 L3= .274 T. and n in rpm. Optimum values according to Paynter's relationships: 02 ). Turbine Type: Francis Specific speed:" 55 (English units).250 L3=.r.0 ~ Q:: 0 20.430X9.t curves for Am = .05 =0. P is in hp.17.2 ). from the values 00" 8 =~= and A2 determined in step b: s.430 L2=.40 s ~ ~ . the optimum values suggested by the author give a better transient response.32 s ~ ~ LI =.05 r. and penstock follow: 1.2S.16 .08 ~ Q:: TIME (seconds) (0) i. = 4 T w = 4X 1.137 L4=. Figure 5.40 '~ ~ ~ .32 L 1=.170 L3=.0 60.255 ~ ./ 0.342 0.40 LI =.16 .
Computations were done as follows: 1.5 = 4. _ ( 127.7.000/0.54.36) 131 579 )1. A (m2) 2:.500 )1.148 Applied Hydraulic Transients WR2 X N2 Hydraulic Transients in Hydroelectric Power Plants 149 Rated turbine output: 127.93 s 106 X 127. (128.9X 7.39 Gg m2 Total WR2 = 4.25 .5 MW Rated head: 74. diameter. and wall thickness for various sections of the penstock are listed in Section 2.3 0. L A excluding draft tube = 9.75s 3.579 Scroll case Draft tube 14.33) = 48.4 4 ~(5.25 T w =!LJ. Experience curves.6 12.5 15.54 9.4 X 106 ~W 4.71)2 = 35.44 Gg m2 kW r ~. Generator Rated output: 125 MW 3.8 244 36.1 = 82. Water starting time.89 1.= A l:  L 10. Since there is a strong possibility of this power plant being isolated from the system.51 0. The length. the generator inertia was selected such that the units would be stable in isolated operation. Penstock.55)2 = 24.7 Normal generator WR )1.5 2.39 = 4.8 ~(6. T w Length. 5.5 = 0.3=35.091 0.970 ( N/ kva .28 Total length ofscroll case = 29 m. Mechanical starting time. MW X kva=Power factor kva= 125.1446 {128. L Conduit Intake Penstock (m) Crosssectional Area.6)1. the following empirical relationships and curves were used.6)2 90AX = 6.7 + 14.5 9.95 m 2.45 Turbine WR2 = 1446 ( N 1.6 X 5.6 m Synchronous speed: 128.55)2 = 24.81 0.6 rpm Flow at rated conditions: 191 m3/s Runnerthroat diameter: 4.5 (Eq.!=_ gHr A 191 X 10.95 2 103 = 131.25 = 15970 ( ' .44 + 0. The diameter of the penstock was determined from an economic analysis so that the incremental benefits from the decreased head losses were more than the increase in the penstock costs.83 Gg m2 . For this purpose.357 6.6)1.83 X 106 X (128.5 )1.1 X 9. L A Remarks (m1) 7. Tm Tm = 90.2 4 ~(5.5(18.81 X 74.6 =2.8 4.88)2 = 18.2 4 ~(4.25 = 15.60 0.2 13.7 4 0.
48 s 38. the values of Tm and Tw were computed above as 6.5 s.2 and 8.  I1H Hr L L 0. Tg = Tc + 1. Waterhammer wave velocities in the penstock were computed in Section 2.48 0. is Tc (s) 6 8 10 Tm = 6. Speed rise. 2. and 8. allowing 1 sec for the cushioning stroke. speed rise for fullload rejection was computed for various values of Tc and T m' The computed values are listed in Table 5 A. The values of Tc for WR2 = 7.6 T m = 11.6 42.27 . Gordon curves. Then.12 =3.9 47.4.22 m2 6.3 47 51. percent 151 a. Waterhammer pressure.6 and 10. in addition to the normal WR1 of 4. Tm 6.6 = 2. were considered.0.6 The waterhammer pressures for various values of Tc were computed from the charts presented in Appendix A.46 s Ae=L L 1:A 295 9. 12 (p. The values of T m for WR2 of 7.3 and 11.8 5 Tm = 10.75 s. p = 2gH = 730X 6. which would result in stable isolated operation.81 X 74. Speed rise. were determined from the intersection of the other two curves. the curve for 4. Water starting time. 14 As both the USBR criteria and the TV A experience curves indicated that the units would be unstable in isolated operation with the normal WR1 . the unit would be unstable in isolated operation.4 63. Wave velocity and crosssectional area for an equivalent 295mlong (including half the length of scroll case) pipe23 was computed as follows: =1:ae a = 31. Of these three curves. Bureau of Reclamation criteria.8 91. excluding that for the draft tube. The computed values are as follows: Tc 6 8 10 Pressure rise.81 aeq 730 .0 Gg m2 are 10.7.2 s. Speed rise.75 As this ratio is less than 8.0 Gg m2.83 Gg m2 did not intersect the curve dividing the stable regions for the isolated and system operation.8 Therefore. Using the procedure outlined in Ref.36 0.2 and 8 Gg m1 .81 X 74. Tw 191 X 9.12 mls 191 V = = o 31.2.93 ==2. in which Tc = effective gateclosing time and Tg = total opening time. Now for different assumed values of T.45 = 9.22 aVo AlIievi's parameter. total WR1 of 7. and for total WR2 equal to 4. These values were 8. T. r. respectively.r? For the normal WR2 of the generator and turbine and for the selected conduit sizes. respectively. b. c.93 and 2. 4.2 and 8.27 s 42. 29).83.S.150 Applied Hydraulic Transients Hydraulic Transients in Hydroelectric Power Plants Table 5. 5.83 Gg m1 .".05 2 X 9. respectively. U. It was found that the units would be unstable in isolated operation. Tennessee Valley A uthority curves. points were plotted on the Gordon's curves. ae = 244 295 51 20 1244 730 mls s ++694 1410 s 2L 295 ==0.93 56. Hence. 13 The values of T m and T w computed in (a) were plotted on the TV A experience curves.0 Gg m2.5. 7.45 Let us assume that the wicketgate opening and closing times are equal.
Use the turbine characteristics shown in Fig.43 Hence.27 s.2 = 7.43X 10. In a loadrejection test on the prototype.18 shows the layout for the Jordan River Power Plant in which a = 0.3 through 5.12. 10. Final check. Proceeding similarly as in Section 5. The chapter was concluded by the presentation of a case study. Tw 2. 0 =~ i\ITm 2.75 0. 10.4 s between the opening of the PRY and the closure of the wicket gates. Various turbine operations that produce the hydraulic transients were discussed.5.2.2 Gg m2 (specified by the turbine manufacturer) Generator WR2 = 7. the wicket gates closed. ii.75 =0. The governor time so that the speed rise following total load rejection does not exceed 60 percent. and the speed deviation following large load changes. The gateopening time such that negative pressures do not occur in the penstock for the minimum forebay water level. the details of the mathematical simulation of the conduit system. a.2 Gg m2 Turbine WR2 = 0.18. the turbine characteristics were available from the model tests conducted by the turbine manufacturer. 5. the values of WR 2 and governor times were selected as follows: a. PROBLEMS 5. maximum speed rise following total load rejection. Develop a mathematical model to analyze the transients caused by a load rejection.75 10. The governor time required for isolated stable governing from Gordon's stability curves.3.75 sand Tm = 10.13 SUMMARY In this chapter. and the PRY opened as shown in Fig. How would the boundary conditions of Problem 5. ii.27 = 0.0 Gg m? Governor closing time = 8 s 7. Develop the boundary conditions for a Francis turbine having a long pres «r w aT m 0. Governor settings. derive the differential equations for this governor. and that there is a delay of 0. 5. optimum governor settings as determined from Fig. 5.. The maximum effective governor time was selected as the minimum of i. The maximum and minimum pressures and speed rise were found to be within the design limits.16 are i\) and surized downstream conduit. 1.27 lOs 8. and the unit was stable following large load changes. 153 6. Prototype test results to verify the mathematical model were presented. The minimum value of the effective governor time is the maximum of i.1. =x. The mathematical model presented in Sections 5. The block diagram for a proportionalintegralderivative (PID) governor is shown in Fig. Figure 3. 5. b.1 be modified if there was a downstream surge tank? 5.4.62 pressureregulating valve (PRY) is provided to reduce the transient pressures.13 r. and governor were outlined.05 X 10. equal to 0. . and compare the computed results with those measured on the prototype (Fig.0.75 == 0.6 was used to check the maximum and minimum transientstate pressures. Selection of WR2 and governor times. Assuming the permanent speed droop. T w = 2. hydraulic turbine. 5.5 X 2. 5.27 2. Based on these criteria.152 Applied Hydraulic Transients Hydraulic Transients in Hydroelectric Power Plants Dashpot time constant.13) Assume that the unit is isolated from the system. the following values were selected: Total WR2 of generator and turbine = 7. During the final design stages.27 = 0. From the preceding computations. Temporary speed droop. equal to 5 percent and the selfregulation constant.5. Procedures for the selection of the generator inertia and for determining the optimum governor settings were then described.19 For these values of i\3 and i\4.3. For the selected conduit sizes and WR2 of the generator. 0.2 . The waterhammer pressure rise following total load rejection does not exceed 50 percent of static head. a.3 = = Tm i\4 == r.
4. E457E472. E. Inc. Internationa/ Association for Hydraulic Research." Jour. P." Proceedings. McGrawHill Book Co. Canada. H. E. of Civil Engineering.. Dept. G. A mer.95 while computing the kva of the unit. 2. 5. May 1516. Fischer.. Determine the unit WR2 required for stable governing of a hydroelectric power plant in isolated operation.. of Mech. "Speed Governor Fundamentals. McGrawHill Book Co... and Darn. 10. 6.. New York.. D.. H.5. 1. Report 71. John Wiley & Sons. Amer. published by British Hydromechanic Research Association. vol.. Rockford. pp. 9. A. L.." Proceeding. Feedback Control System Analysis and Synthesis. Second Edition. Annual Meeting.1973. Inc.. "Hydropower plant Transients. D.7. B. Fluid Mechanics. . Signals. Streeter. E. Zolotov. Chaudhry. 1966.67 . et ai. 298314. Wordward Governor Company. 7.154 Applied Hydraulic Transients Hydraulic Transients in Hydroelectric Power Plants 155 5.. and Klabukov." presented at the spring meeting. Systems and Communications. 1966. 1216112168. New York. W. Paris. Soc. C.6. "Moment Characteristics of Cascades under Nonstationary Flow Conditions. of Civil Engrs. "The Westinghouse Leading Edge Ultrasonic Flow Measurement System.. April 1957. Soc. Portfors... McCracken.. New York. V. and Chaudhry... N. published by the University of Alberta. "A Mathematical Model for Analyzing Hydraulic Transients in a Hydroelectric Powerplant. Massachusetts Inst. pp.. 1971. D'Azzo. BI11B 119.9 m2 Governor opening and closing time = 5 s Neglect the length of the draft tube and assume a power factor of 0. Power Div. 3. Massachusetts. G. and the scroll case at the upstream end = 7. Canterbury. J. H. 5. John Wiley & Sons. and Houpis.lllinois. 4th Edition. Boston. A. First Canadian Hydraulic Conference. "Analysis and Prototype Verification of Hydraulic Transients in Jordan River PowerpJant. M.Y. Krivehenko. 5. Develop the boundary conditions for a Kaplan turbine taking into consideration the variation of the blade angle. of Tech.. Parmakian.. First International Conference on Pressure Surges." Proceedings.. and Portfors... REFERENCES 1.. S." Part II and Ill.. Engineers. Sept. England. A." Bulletin 25031. What are the optimum governor settings for the unit of Problem 5. ct ~ Ii ~ b . J. pp. F. I> c:: '" ~ ~ s . Edmonton. 1964.pp. 8.. 2. J. Sept. New York.'" 0:: 'd:: . 1972. M. M. The data for the power plant follow: Rated output = 39 MW Synchronous speed = 500 rpm Rated head = 240 m Turbine discharge at rated conditions = 38 m3 /s Length of the penstock = 640 m Length of the scroll case = 36 m Crosssectional areas of the penstock. Numerical Methods and Fortran Programming. V. S. J.. 11. Lathi. 1964. Perkins. May 1973. "Water Hammer Design Criteria. Hydrodynamics Lab.1965 .
July Aug. 445455. June 1970.AprilI971. 1965. of Elect. 1974. pp. Stein. Annual General Meeting Paper No. Vaughan. 3. Nov. "Influence of Water Inertia on the Stability of Operation of a Hydroelectric System. Paper No." Water Power. pp. F. uf Mech." Part I and 1I. 1962. pp. 14. Paynter. pp. and Hill. 63WA15. Goldwag. M. International Code of Testing of Speed Governing Systems for Hydraulic Turbines. 22. 323328.... J. L. 912.. "Frequency Response Measurements at Vianden.." Trans. Amer." Eng. (U. 140. pp. "Determination of Generator Inertia. "TieLine Power and Frequency Control of Electric Power Systems." Engineer's Digest. pp. 604605). of Elect. 13. 13. L. ADDITIONAL REFERENCES Alrneras.7. July/Aug.. 333341. Aug. Inst. G. "Frequency Control under Isolated Network Conditions. Advanced Engineering Mathematics. pp. (See also discussion by H. pp. 22. "A Palimpsest on the Electronic Analogue Art... and Electronics Engrs. A mer. Brekke.A). H. and Felt. May 1953.pp. pp. Schleif. pp. "Optimum Adjustment of Hydro Governors on Manitoba Hydro System. 599604. "The Influence of SelfRegulation and of the Damping Period on the WR2 value of Hydroelectric Power Plant. Engrs. Dec. Wylie. Engineering Inst. "Analysis of Governing Stability of Hydroelectric Power Plants. Technical Report No." The Engineers' Digest. Amer. (U.. R. 882890. 1961. Soc. R." Trans. 1970. IV. Basic Engineering. McGrawHili Book Co.. "Extension of Stability Boundaries of a Hydraulic Turbine Generating Unit. 480493. Waterhammer Analysis. M. H. 1965. "Hydraulic Speed Governor with Major Governor Problems Solved. lnst." International Symposium. Basic Engineering. Soc." Water Power. 1947. "WaterTurbine Governors and the Stability of Hydroelectric Plant. Engrs." The Engineering Journal. Boston. 6576. J.4. "Field Testing and Simulation of Hydraulic Turbine Governor Performance. 17. pp.. C. M. vol. Inst . Hovey. Stein. Amer. "Speed Regulation Study for Bay D'Espoir Hydroelectric Generating Station." Water Power. T. Amer. International Electrotechnical Commission. Sept.. January 1961. p. Feb.. "Optimum Adjustment of Governors in Hydro Generating Stations. Thorne. Mass. E. M. R. pp. Dover Publications. 1961." Trans. of Mech." presented to the Canadian Electrical Association. (Translated from Schweizerische Bauzeitung .. L. pp. London. Philbrick Researches. R. 581586. "On the Influence of Water Turbine Characteristic on Stability and Response. Inc. 1947.. Inst.. England. "Mechanical Design of Hydro Plants. A. 5055. T. Dec. Feb. F. M. Inst. "The Use of an Electronic Analog Computer for Determining Optimum Settings of Speed Governors for Hydro Generating Units... 562572. (See also discussion by Chaudhry. 1967. pp. Jan.. 1947. M. "Conduit Representation in Closed Loop Simulation of Hydroelectric Systems. 1953. D. Engrs. M. pp. vol. of Canada. 104109. pp. MayJune 1948.. 191196." Jour. A mer.. 1953. Engrs." Tennessee Valley Authority Projects. of Canada.S. vol. 18. 1955. July 1967. 18.Oct." Trans.. and Electronics Engrs. 1964. Dennis. 721. 1972. E. of Mech.. L.) 19. Sept . of Canada.. April 1977. pp. Hovey. Paynter. 1967. Engrs. R. P. of Mech. M. Chaudhry. L. Amer. November 1960. PAS94. of Elect. M. Hovey." Water Power. and Ruus.. pp. 1971. 1975. Chaudhry. 2. "Regulation of a Hydraulic Turbine Calculated by StepbyStep Method. Apr. Soc. Halifax. D. H. Araki. Lein. pp. 1963.. E. H. 1962." thesis submitted to Massachusetts Institute of Technology in partial fulfillment of the requirements for the degree of doctor of philosophy. April 1954. April 1970. Sept. pp. A mer. A mer. June 1953. Sept. Basic Engineering. C. lnst . K. 15. InSI. 1966. D." Jour." Water Power." Hitachi Review. 14011409. 20. G. Engrs. 647 1.. I. H. 5561. Mar. of Elect. 20." Trans.. T." Amer.. Power Apparatus and Systems. 151154. G. 67WA/FE17. and Bates.. of Elect. M. Mech." Engineering Monograph No." A. H... . Newey. London. 364368. 23. Basic Engineering. Soc. Mar. "Governing Stability of a Hydroelectric Power Plant. Technical Committee No. H. Sept. "Speed Regulation Tests on a Hydrostation Supplying an Isolated Load. E. Trans. 11831191. Engrs. pp. pp. Engrs. Concordia. no." Trans. Dec. 1974. K. Engineering Inst. L. L.. F. R. 4. pp. 283286. 21." Jour. and Kirchrnayer... 131136. 1960.. Parrnakian. presented at Winter Annual Meeting." Jour.. Hydraulic Turbines. of Mech.A). Sue.156 Applied Hydraulic Transients Hydraulic Transients in Hydroelectric Power Plants 157 12.. Feb. Krueger. of Elect. Engrs. 24. Wozniak. "Speed Control in Hydroelectric Power Systems. Thorne. "Governing Characteristics for 820.. 3rd Edition. C. of Elect. 114." Paper No. pp. E. Soc. and Kuwabara.. vol.000 Horsepower Units for Grand Coulee Third Powerplant. 16. Power Apparatus and Systems.. G. and Hill. New York. pp. 1962. Inst. "IEEE Recommended Practice for Preparation of Equipment Specifications for Speed Governing of Hydraulic Turbines Intended to Drive Electric Generators. N. and Electronics Engrs.) Sweiecicki. "Selecting Hydraulic Reaction Turbines. Oldenburger . "Waterhammer in Pumped Storage Projects.. "Water Column Effect on Speed Control of Hydraulic Turbines and Governor Improvement. and Parzany. Gordon. Hovey. New York. "Stability Studies for a Governed Turbine Operating under Isolated Load Conditions. H. and Electronics Engrs. United States Bureau of Reclamation.S. vol... 1953..
oclor coolonl pump Figure 6. In this reactor..rolor R. A pressurizer? is provided in this loop to control the pressure of the coolant.g. helium. Pressurized water reactor (PWR). boiling ordinary water). generated in the nuclear reactor by splitting the atoms. Different methods for analysis are discussed. It is necessary for the design and for the safe operation of these systems that the transientstate conditions caused by various operating conditions be accurately known. The reactor coolant may be a liquid (e. Thermal energy. Therefore. (e.Hydraulic Transients in Nuclear Power Plants 159 CHAPTER 6 HYDRAULIC TRANSIENTS IN NUCLEAR POWER PLANTS coolonlpump Figure 6.2 TERMINOLOGY General Figure 6.2. heavy water [D20] . ordinary water is used as a reactor coolant and as a moderator. carbon dioxide [C02]).1. suitable for the analysis of twophase transient flows. water is 6. 158 . The temperature of the coolant rises as it flows through the reactor vessel. 6.. used to slow down or moderate the highenergy neutrons produced by the fission process.2). This loop is called the primary loop. Schematic arrangemen t of a nuclear power plan t..g. Water coolant is circulated at high pressure (about 14 MPa) by the external pumps through the reactor vessel. and special boundary conditions for the method of characteristics are derived. Water flows upward through the fuel clusters. are then presented. and the pipe on the exit side is called the hot leg (Fig.g. the pipe on the inlet side of the reactor is called the cold leg. Reactors are classified I according to the type of fuel. various piping systems are used for cooling and for heat transfer. ~SI.om VSI.r 10 lurbine (I~n.1 shows the schematic arrangement of a nuclear power plant. terminology is introduced first. Types of Reactors 6. Three main types of liquidcooled reactors are: 1. various operations which may produce transients are then outlined.om LLF"'dwol.1 INTRODUCTION In nuclear power plants. ordinary water. reactor coolant. it may be a liquidvapor mixture. Pressurized water reactor. liquid sodium). Since various types of reactors are available and each reactor type differs in details from one manufacturer to another. and the type of medium. In the secondary loop. is transferred from the reactor by a fluid medium called the reactor coolant. called moderator. or it may be a gas (e . The details of a numerical scheme. the analysis techniques are emphasized instead of the specifics of design details. In this chapter. then from the vessel into the heat exchangers and back to the pumps.
160
Applied Hydraulic Transients
Hydraulic Transients in Nuclear Power Plants
Inltlrmtldioltl IItlol tlJfcllongtlr
161
boiled in the heat exchangers forming saturated steam, which is used to drive the steam turbine. 2. Boiling water reactor (BWR). This reactor is similar to the PWR except that ordinary (light) water coolant is permitted to boil in the reactor core. The steam thus produced is separated from the coolant by centrifugal separators (Fig. 6.3) located in the reactor vessel above the core. This steam is then directly fed into the turbine. 3. Liquidmetal fast breeder reactor (LMFBR). In this reactor, liquid sodium I is used as the reactor coolant, which is cooled in the intermediate heat exchangers (Fig. 6.4) and is returned to the reactor. The intermediate heat exchangers are cooled by a second flow of liquid sodium, which in turn is cooled in a second set of heat exchangers in which steam is produced for the turbine. The steam exhausted from the turbine is condensed by means of a condenser, and the condensed water is pumped back into the heat exchangers. A large amount of water has to be pumped from a lake, river, or ocean to the condenser loop (Fig. 6.1) for this purpose.
Sltlom 10 lurlJintl
Sltlom gtlntlrolor Rtloclor cor« Ftltldwoltlr
coolonl pump
Inltlrmtldioltl circuloling pump
Figure 6.4. Liquidmetal fastbreeder reactor,
Emergency Core Cooling Systems The emergency core cooling (ECC) systems? are used to provide coolant for possible lossofcoolant accident (LOCA). A number of subsystems are employed for this purpose: 1. Highpressure coolant injection system. This system employs highhead, lowcapacity pumps and is intended to provide coolant during smallsize breaks
Rtloclor "tlsstll Sltlom 10 lurlJintl Sltlom woltlr stlporolor
in the primary loop. This system is activated during the early stages of depressurization, 2. Lowpressure coolant injection system. This system employs lowhead, highcapacity pumps; it is intended for supplying coolant following largesize breaks in the primary loop and provides coolant during later stages of LOCA. 3. Accumulator System. This is a passive system and is intended to provide large volumes for largesize breaks. A check valve, which isolates the accumulator from the reactor system during normal operation, opens whenever the pressure in the reactor system drops below the pressure of the accumulator and provides coolant to the primary loop. 6.3 CAUSES OF TRANSIENTS Transients in the piping systems of a nuclear power plant are caused by the following (only events initiated in the hydraulic system are consideredj.!"!" 1. 2. 3. 4. 5. 6. Planned or accidental starting or stopping of pumps; power failure to pumps Planned or accidental opening or closing of the control valves Instability of pumps Release of entrapped air or collapse of vapor bubbles Wave action at the reservoir water surface (for the condenser piping system) Rupture of pipeline.
~t'., j.::::::._=::.::::,h,
Ftltldwoltlr
Figure 6.3. Boiling water reactor,
The starting or stopping of the pumps, the opening or closing of control valves under controlled conditions, and power failure to the pumps are con
162
Applied Hydraulic Transients
Hydraulic Transients in Nuclear Power Plants
163
sidered normal operations. Flow variations may be caused by pump instability due to abnormalities in the pump characteristics. If the pumps are started into an empty pipeline or into a partly drained discharge line, severe pressure oscillations may be produced. Waterhammer pressure generated by air release or due to collapse of vapor or steam bubbles has resulted in damage to the pipelines of a number of nuclear power plants.'?: 14 Rupture of a pipe in the primary 100p2 and the loss of reactor coolant, i.e., lossofcoolant accident, is one of the worst conditions to be considered in the design. Rupture of, or a leak in, the pipe in the secondary loop of a LMFBR will result in sodiumwater reaction; transients produced by such an incident have to be considered in the design stages. 6.4 METHODS OF ANALYSIS General Remarks The following two approaches are available for the analysis of hydraulic transients in the piping systems of nuclear power plants: I. Lumpedsystem approach 2. Distributedsystem approach. In a lumpedsystem approach, sometimes referred to as controlvolume or macroscopic approach.P the system is divided into a number of con trol volumes, and an integrated form of the continuity and momentum equations (and the energy equation, if required) is used. Thus, there is no continuous spatial variation (i.e., with respect to distance x in a onedimensional model) of various variables, and they vary with respect to time only. Therefore, a system of ordinary differential equations, instead of partial differential equations, describe the transients.l" Since the initial steadystate conditions are specified, the transient analysis using a lumpedsystem approach may be considered as an initial value problem. In a distributedsystem approach, both the temporal and spatial variations of various variables are taken into consideration. Thus, a system of partial differential equations subject to appropriate boundary conditions describe the behavior of the system. If there is simultaneous flow of two phases (e .g., water and vapor, air and water), then the flow is called a twophase flow. 17 Twophase flows may be further classified as homogeneous or separated flows. We will consider only the homogenous twophase flows in which the mixture of the two phases may be treated as a pseudofluid. The separated flows are beyond the scope of this book.
If heat is added or removed from the system during a transient, then the energy equation has to be considered in addition to the continuity and momentum eq uations. Formulation of Mathematical Models Extreme caution must be exercised in formulating a mathematical model since a model based on questionable simplifying assumptions may yield totally incorrect results no matter how sophisticated are the numerical techniques employed to solve the governing equations. Similarly, using too complex a model, where a simplified model would suffice, results in waste of manpower as well as computer time. A lumpedsystem onedimensional model is one of the simplest, whereas a multidimensional, distributedsystem, twophase model with heat addition or removal is the most complicated. The following factors are considered while selecting a model for the analysis of a particular system: 1. For a slow transient phenomenon, the compressibility effects may be neglected and the system may be analyzed by using a lumpedsystem approach. This approach may be used18 for systems in which wl/a« I (see p. 219). In this expression w = forcing frequency; I = length of the pipe, and a = wave velocity. As a rough rule of thumb, wi/a = 0.05 may be considered as the upper limit for the validity of the lumpedsystem approach. 2. If the void fraction, a (a = volume of the gas and vapors per total volume of the gasvaporliquid mixture), is small, then its effects can be totally neglected or may be taken into consideration by reducing the waterhammer wave velocity. 3. Flows having a large void fraction, have to be analyzed as twophase flows. If the phases are not separated, then a homogeneous flow model may be used in which the gasvaporliquid mixture is treated as a pseudofluid and the continuity, momentum, and energy equations for a singlecomponent flow are used in the analysis. However, if the phases are separated, then the continuity, momentum, and energy equation for each phase have to be used, and the transfer of mass, momentum, and energy between the phases has to be considered. Such a model is called a separated flow model. 4. If there is significant heat input or output from the system during the transient conditions under consideration, then it is necessary to include the energy equation in addition to the continuity and momentum equations in the analysis. Numerical Solution The numerical method to be used for the solution of the equations describing the system behavior depends upon whether a lumpedsystem or a distributed
164
Applied
Hydraulic
Transients Various available methods are discussed in the
':'_. 'lfy"drauIic~TtatiSiertts
In Nuclear Power Plants
165
system approach is being used. following paragraphs.
LumpedSystem
Approach
As discussed previously, a mathematical model based on a lumpedsystem approach is comprised of a system of ordinary differential equations. A number of finitedifference methods 1920 are available to solve these equations. The author's experience with the solution of these equations indicates that the fourthorder RungeKutta method20•21 is quite versatile and accurate. In addition, most computer installations have standard packages available for this method.
DistributedSystem
Approach
subdivided into two categories, or twophase:
In the method of characteristics,2225 the discontinuities in the derivatives can be handled, and the boundary conditions are properly posed. The method, however, fails because of the convergence of characteristics curves if the wave velocity is highly dependent upon pressure, and a shock is formed in the solution. In addition, if an explicit finitedifference scheme is used to solve the total differential equations obtained by this method, the CourantFredrichLevy conditlon=' for the stability of the numerical scheme has to be satisfied. This condition requires the use of small computational time steps, thus making the method unsuitable for solving reallife large systems. The method may, however, be used to verify other numerical schemes by analyzing small, simple systems. The LaxWendroff twostep finitedifference scherne i" is the most suitable for analyzing systems in which a shock forms. However, the scheme produces oscillatory solution behind the wave front, and a smoothening parameter'" has to be introduced to avoid this. This introduces numerical damping, which is not present in the actual system and which, if not properly taken care of, may smoothen the transient peaks. Because the Courant's stability condition has to be satisfied, the size of time steps is restricted, which makes the scheme uneconomical for general analyses. Explicit finitedifference methods are very easy to program. However, as the step size is limited by the Courant's stability condition;" a large amount of computer time is required. Thus, the method is not suitable for analyzing large systems. In the implicit finitedifference methods, the size of the time step is governed by accuracy only and not by the stability considerations.2932 These methods are therefore useful for the analysis of large systems. Details of this method are presented in Section 6.7.
Distributedsystem approaches may be further depending upon whether the flow is singlephase
I. Singlephase flows. The method of characteristics presented in Chapter 3 may be used for singlephase flows. To account for a small amount of gaseous phase in the liquid, a reduced value of the waterhammer wave velocity (see
Section 9.5) may be used in the analysis. The transients caused by opening or closing of valves, by starting or stopping of pumps, or by power failure to the pumpmotors may be analyzed by using the method of characteristics. A number of commonly used boundary conditions were derived in Chapters 3 and 4, and a few additional ones are developed in the next section and in Chapter 10. 2. Twophase flows. The twophase flows may be analyzed by considering them as homogeneous or separated flows. In the case of homogeneous flow, the liquid mixture is treated as a pseudofluid, and the averaged values of the various variables (such as pressure, flow velocity, and void fraction) over a cross section are used. The spatial variation of void fraction may be included in the analysis. In the analysis of separated flows, each phase is treated separately, and the transfer of mass, momentum, and energy between each phase is taken into consideration. Analyses of these flows are beyond the scope of this book. The following numerical methods have been used for the analysis of homogeneous twophase flows: 1. Method of characteristics 2. LaxWendroffs twostep finitedifference 3. Explicit finitedifference methods 4. Implicit finitedifference methods.
6.5 BOUNDARY
CONDITIONS
To analyze a piping system by the method of characteristics, the boundary conditions should be known. A number of boundary conditions commonly found in the piping systems of nuclear power plants are derived in this section. Note that these conditions are valid only for singlephase flows and are required if the method of characteristics of Chapter 3 is used for the analysis. Condenser
method
A condenser is comprised of a large number of tubes with water boxes on the ends (Fig. 6.5). To derive the boundary condition, the cluster of tubes may be replaced by an equivalent pipe having a crosssectional area, Ae, equal to the
166
Applied
Hydraulic
Transients
I:
Hydraulic
on enser f ub e, A t
Transients
in Nuclear Power Plants
167
Wafer bO/lVolume Pipe i
in which A V is the change in volume due to change in pressure, ap. For the pressure changes usually encountered in practice, A V is small, and therefore V may be assumed constant. The change in volume, AV, during a time step At, may be determined from the con tinuity equa tion
¥
A¥::
! M [(Qpi,
tI+1 + Qi, n+1) 
(QPi+I,1
+
Qi+1.1)]
(6.2)
C~.
(I, n+l)
in which Q and Qp are discharges at the beginning and at the end of the time step, and subscripts (i, n + 1) and (i + 1, 1) refer to the section numbers (see Fig. 6.5). If it is assumed that the pressure is same throughout the box, then
(6.3)
in which H P = piezometric head above the datum at the end of lime step. Now,
£iqU/l/aIen .
r
Ap
pIpe,
= "(AH
= "(Hp.
I,
n+ 1 
Hi , n+l)
(6.4)
the resulting
Ae= nt At
in which "( = specific weight of water. By substituting Eqs. 6.2 and 6.4 into equation, we obtain
Eq. 6.1 and simplifying
Wafer bO/lVolume Pipe i
"'__Wafer bO/l
,
(i+ I, I) Pipe i+2
¥
KAt Hpi.n+1 =Hi.tI+1 + 2yJ.! [(Qi,n+1  Qi+1,d+
The positive and negative characteristic equations
(QPi.Il+1  QPi+1,)]
(6.5)
(I.
,
(Eqs. 3.18, 3.19) for sections
(i, n
Figure 6.5. Condenser.
+
1) and (i + I, 1) are
(I, n+ I)
QPj.
n+1 =
Cp

CajHpj, n+1
(6.6) (6.7)
3.22. Substitution
Qpi+I,1
=Cn+Cai+1HPj+1,1
combined area of all the tubes, i.e., Ae = ntA t, in which At = crosssectional area of one tube and nt = number of tubes in the condenser (Fig. 6.5). The head loss in the equivalent pipe is, however, assumed equal to the head loss in an individual tube. The water boxes may be considered as lumped capacitances, and the compressibility of water and the elasticity of the walls of the boxes may be taken into consideration. Equations for the upstream water box are derived below; equations for the downstream box may be derived in a similar manner. Let the volume of water in the box be ¥, and let the combined effective bulk modulus of water inside the box and the vessel walls be K. Then, by definition,
in which Cp, Cn, and Ca are as defined by Eqs. 3.20 through of Eqs, 6.3, 6.6, and 6.7 into Eq. 6.5 yields
Hpj,n+1 =Hi,n+l
+ 2r¥ [(Qi,n+l  Qi+l,l)+
KAt  2"(¥ (Cai + Cai+1 )Hpi,
n+1
KM
(Cp

Cn)] (6.8)
Hence,
HPj,n+1 = 2'}'¥+KAt(C
2y¥
K=
Ap
a,
.+C.)
{ a,+1
H· + KAt¥[(Q"n+1 . ',n+1 2 y
. Q,+I,I)
+ (Cp  Cn)]} f6.9)
A¥ ¥
(6.1)
168
Applied Hydraulic Transients
QPi+l.l
Hydraulic Transients in Nuclear Power Plants may be determined from Eqs. 6.3, 6.6,
169
Now, Hpi+I,1 ' Qpi, 11+1' and and 6.7, respectively. Entrapped Air
Now we have six equations (Eqs. 6.106.15) in six unknowns, namely,Hp* . , air and Hp. 10 n +1' Elimination of the first five unknowns from these equations yields
¥p au , Qp. I, n +1' Qp. 1+ 1• I' Hp. 1+ 1, I' .
(6.16) in which (6.17) Equation 6.16 may be solved for Hpi 11+1 by an iterative technique, such as the bisection method.i? The values of th~ other unknowns may then be determined from Eqs. 6.1 0 through 6.15. Pipe Rupture and failure of Rupture Discs Sometimes, rupture discs are installed, which fail if the pressure inside the pipeline exceeds a specified limit. Because of this controlled failure, extensive damage to the pipeline is avoided. The ruptured discs or pipebreak may be analyzed as an orifice, and the boundary conditions derived in Chapter 3 may be used for this purpose. If there is back pressure from outside the pipeline, then t.H is the difference between the pressure inside and outside the pipeline. Total pipebreak may be analyzed considering it as a fixedopening valve located at the ends of pipe at the location of the pipebreak. 6.6 LOSSOfCOOLANT ACCIDENT
Let us consider a volume of air en trapped in a pipe having liquid on either side as shown in Fig. 6.6. If the entrapped air follows the polytropic law for perfect gases, then (6.1 0) in which and ¥p. air are the absolute pressure head and volume of the enau trapped air, respectively, and m = exponent in the polytropic gas law. The value of the constant, C, in Eq. 6.10 is determined from the initial steadystate conditions. From the continuity equation, the following equation may be written for the volume of the air:
¥Pair=¥air+t t.t{(Qi+I,1 +QPi+I,I)(Qi,n+1 + Qpi,n+l)}
H;.
(6.11)
The positive and negative characteristic sections (i, n + 1) and (i + 1,1) are
Qpi,n+1 Qpi+I,1 =Cp =
equations
(Eqs, 3.18, and 3.19) for (6.12) (6.13)
 CaiHpi,n+1 Hpi+1 ,I
c; + Cai+1
in which Cp, Cn, Cai, and Cai+1 are as defined by Eqs. 3.20 through 3.22. If the pressure of air at any instant is assumed to be same throughout its volume, then (6.14) In addition,
H*. =H +H . Z Pair b P"n+1
(6.15)
in which H b
=
barometric head, and z
= height
of the pipeline above datum.
Air Pip,
..•.•.•••.•.•.•..•.•.•..... m···.·;I·:tI·.IIItl
( i, n+l )
Figure 6.6. Entrapped air.
Pip'
i +1
( i+ I, I)
A sudden rupture of a pipe in the primary loop and the resulting loss of reactor coolant is referred to as lossofcoolant accident. The size and location of the pipebreak in this hypothetical accident is selected, which results in the maximum cladding temperatures. In a PWR, a complete rupture of the pipe connecting the pump to the reactor (i.e., cold leg) is assumed." In a BWR, a complete and instantaneous circumferential rupture is assumed? of the largest pipe in one of the suction lines of the recirculation system. The analysis of LOCA is done by using a number of mathematical models, and the output of one model serves as input to the other model. For example, the average conditions in the core, such as flow, pressure, and temperature, are determined by a mathematical model of the primary loop. Using these computed conditions as input, the maximum cladding temperature is determined using another model. These two models are used in an iterative manner at every time step or after a number of time steps. In the mathematical model of the primary loop, the continuity, dynamic, and
= void fraction = volume of vapor per total volume of the vaporliquid mixture.29) (6. and the subscripts g and 1 refer to the vapor and the liquid. Continuity equation ap 2 av + pa + at ax ap V=c ax 3 (6.20) In these equations. In this method.30 the homogeneous. the details of an implicit finitedifference method.23) (6. f = friction factor.19) P = o. T w = wall temperature. p = density of the fluid. pipes including the heat addition or loss: 1. F:: friction force per unit mass. the containment vessel is decoupled by specifying a back pressure at the hypothetical pipebreak. A = crosssectional area of the pipe.21) = a2 [(q + VF) ap = a2 [(q + h I h . g = acceleration due to gravity. Therefore.g cos e ap (6. x' = thermodynamic quality. The heat addition in the core to the fluid of the primary loop and the heat loss in the heat exchangers from the primary loop to the secondary loop are assumed to be distributed along the pipe lengths representing the core and the heat exchangers. T* T* = 1 for x' <0 and x' >1 < x' < 1 > 2000 < 2000 } } = appropriate twophase multiplier for 0 for RN for RN (6. kw = wallheat transfer coefficient. a larger size of time steps can be used. A numerical method is presented in the next section to solve the equations describing the flow conditions in the primary loop following a pipebreak. h = mixture enthalpy. p = pressure. kif = distributed loss coefficient.0. Energy equation +a2 ah at + V::c av ax ax ah h =x'hg + (1 . However.1 (6.046RNo. RN = Reynolds number.24) q :: kwAw (T w p F T) t (6. =F . 30. de = equivalent pipe diameter.27) + av at ap V+::Cl ax p ax av I f= 0.26) and in which a = wave velocity in the fluid. the method is as similar as possible to the method of characteristics.25) = (4f d T* + ~ ) VWI I 2 (6. V = fluid velocity. presented by Hancox et a1. the characteristic form of the equations governing the homogeneous twophase flows are used. as it is not necessary to satisfy the Courant's stability condition. respectively.18) t= 64 RN (6.) PI 3. Momentum equation twophase flow in a=Cpl op +_!_~I p 3h  Ip r I + pV p l12 1 (6.V dA] A dx dA A dx (6.28) 2. Governing Equations The following equations describe28. 31 are outlined. Similarly. thus making the method suitable for the analysis of large systems. Tf = temperature of the fluid. The secondary loop is considered as a heat sink and is therefore decoupled from the primary loop. .22) C3 VF) ap 3h 6.170 Applied Hydraulic Transients in which c1 C2 Hydraulic Transients in Nuclear Power Plants 171 energy equations are solved.30) (6. In addition. 0.7 IMPLICIT FINITEDIFFERENCE METHOD FOR ANALYZING TWOPHASE TRANSIENT FLOWS General In this section. x = distance along the pipe axis.28. and e = angle between the pipe axis and horizontal. t = time.x') hI 2 (6.Pg + (l . q = walltoflow heat transfer.
These equations can be converted into ordinary differential equations by the method of characteristics as follows: Multiplying Eq . Eq. it can be proved that Eqs. pdh . Note that. 6.41 are .pac! (6. 6. 6.19. and rearranging the terms.18 by a linear multiplier.34 is valid only if Eq.43) v av] + rap +(Va)ax at ap] =cJpac ax I (6.32 yields +(V+a) :~]=c3+pacl (6. These equations can be combined into matrix form as B+ at au C=D ax au (6.e. Similarly. multiplying Eq.36.20 form a set of hyperbolic partial differential equations.in the socalled characteristic form. 6. =C2 . 6.41 is valid along the path of a particle in the xt plane. Now. pa dV + dp 2.37) padV+dp (6. 6.dp = PC2 C3 [6.ng characteristic curve. and rearranging the terms.44) = V.42) Equation 6. Equations 6. we obtain AI[aa~ + Let pal dx Vt==V+___!. Thus. unlike Eqs. 6.38 yields A p (6.172 Applied Hydraulic Transients Hydraulic Transients in Nuclear Power Plants equation to Eq.34) dx =V dt (6. Along dx/dt = V . ~'1' adding the resulting equation to Eq. 6.436.31) (6. 6.a. Along dx/dt = V + a. dxldt = V + a. Equations 6. and 6.46) .33) +(V+a) ~:]+[:~ =r oa into Eqs.41 can be written as: 1.32) and p p ah + V ah) _ _!_(a + V a ) = C2 ( at ax p at ax  _!_C3 p (6.45) Note that Eq. Eq. Eq. 6. and 6.6. Along dxldt (6.34 is valid alo.41) or AI = ±pa Substitution of AI =paandAI pa[~~ if dx = V+a dt and oa [a +(Va)a t if dx = Va df (6.36) 3. 6.. in the xt plane.35) = C3 + oac.20 by a linear multiplier (Al).45) are called compatibility equations. 6.40 into Eq.41 is valid if Eq.39) or I A2 =P (v+ P:ll)~:]+[:~ (v+ + ~I):~ ]=c 3 + Aici (6. which were valid along the characteristic curves dxldt = V ± a.42 is satisfied.31 and 6.36 is valid along the characteristic curve dx/dt = Va.34 and 6.40) Substitution of Eq.35 is satisfied. AI dt (6.4.20.6. we obtain ah ( +Vat Let Oh) +A2 (oP +Vox at OP) +al(l+pA2)=A2CJ+C2 oV ax ox J 73 Conversion of Governing Equations into Characteristic Form By following the procedure outlined in Section 2. i. 6. 6.18 through 6. 6.34. adding the resulting These ordinary differential equations (Eqs.36. (6.34.36.38) (6. 6.
7.+1 .49) j+I L. 6. = D{ At I Ax.+l . the lefthand finite differences are used for the positive characteristic (V + a equation).Ui. Eq. For subsonic flows. depends upon the flow direction and whether the flow is subsonic or supersonic.7) in Eq.' .50) 1 I I I .a equation). . 6.174 Applied Hydraulic Transients Hydraulic Transients in Nuclear Power Plants 175 in which (6. in the Vequation.~ I I " I I II I (6..or lefthand finite differences are used depending upon the flow direction.52) r.Ui. At ' AX.' . I .46 can be expressed in the finitedifference form as U~+I .~l\ .51 ) j+1 (6. '' + Ci. (6. C3  paci iI 1+1 Positive V (b) V Equation (I) iI (2) Negative 1+1 V Formulation of Algebraic Equations Referring to Fig. . 6.U~ U~+1 .~) I I I r= I f' I I I Note that the spatial derivatives are approximated by the lefthand finite differences (Fig. U/+1 .7. and the righthand finite differences for the negative characteristic (V .51 and by the righthand finite differences in Eq._I Bi. iI (e) (Va) 1+1 Equation Figure 6. + Ci '+1 .U~+1 Bi. the right. This convention makes the finitedifference scheme as similar as possible to the method of characteristics. (i_I j+1 D= { C3 +~aCI} C3 C2  1 1 I I I I I r :: : I I P (6. Ui.48) iI (a) (V+a) 1+1 Equation pa(V+a) c= [ 0 pa(V . Notation for implicit finitedifference scheme (subsonic flow).l"\ .I = D{ . Which of these two equations should be used. 6.47) j+1 .a) (6. 6.52.
This action might not be possible to undo. Are you sure you want to continue?
We've moved you to where you read on your other device.
Get the full title to continue listening from where you left off, or restart the preview.