ELECTRICAL TRANSIENTS IN POWER SYSTEMS Second Edition Allan Greenwood Rensselaer Polytechnic Institute Electric Power Engineering Department Troy, New York feeoru oF ee ECDACA A WILEY-INTERSCIENCE PUBLICATION BIBLIOTECA JOHN WILEY & SONS, INC. New York © Chichester * Brisbane * ‘Yoronto * SingaporeCONTENTS Preface to the First Edition xiii Preface 1 Fundamental Notions about Electrical Transients 1 11 12 1.3 14 15 Introduction 1 Circuit Parameters 1 Mathematical Statement of the Problem and Its Physical Interpretation 3 Circuit Characteristics or Thumbprints 5 The Principle of Superposition 6 Problems 9 2 The Laplace Transform Method of Solving Differential Equations ul 21 2.2 2.3 The Concept of a Transform 11 The Laplace Transform 12 Some Simple Applications of the Laplace Transform in Circuit Problems 18 Building Other Transforms 26 Operational Impedance 27 Duhamel’s Integral—Response of a Circuit to an Arbitrary Stimulus 29 Problems 33 References 36 3 Simpie Switching Tansients 37 3.1 3.2 3.3 3.4 Introduction 37 The Circuit Closing Transient 37 The Recovery Transient Initiated by the Removal of a Short Circuit 47 Double-Frequency Transients 52 Problems 57 References 61 4 Damping 62 4.1 Some Observations on the RLC Circuit 62viii CONTENTS 4.2 43 44 4.5 4.6 4.7 4.8 The Basic Transforms of the RLC Circuits 64 The Generalized Damping Curves 72 The Series RLC Circuit 75 Resistance Switching 80 Load Switching 83 Other Forms of Damping 87 Damping and Frequency 87 Problems 88 References 91 Abnormal Switching Transients 92 SA 5.2 5.3 5.4 5.5 5.6 Normal and Abnormal Switching Transients 92 Current Suppression 92 Capacitance Switching 100 Other Restriking Phenomena 104 Transformer Magnetizing Inrush Current 113 Ferroresonance 116 Problems 122 References 124 Transients in Three-Phase Circuits 126 61 6.2 63 6.4 6.5 Introduction 126 Importance of the Type of Neutral Connection 126 Switching a Three-Phase Reactor with an Isolated Neutral 128 Three-Phase Capacitance Switching 132 The Symmetrical-Component Method for Solving Three- Phase Switching Transients 140 Problems 147 References 149 Transients in Direct Current Circuits, Conversion Equipment and Static Var Controls 150 7.4 Introduction 150 7.2 Interruption of Direct Current in Low Voltage Circuits 150 7.3 Transients Associated with HVDC Circuit Breakers 158 7.4 Delayed and Periodic Functions 160 7.5 Characteristics of the Thyristor and the Gate Turn Off (GTO) Thyristor 166 7.6 Commutation Transients—The Current-Limiting Static Circuit Breaker 169 7.7. Commutation Transients in Conversion Equipment 17410 CONTENTS = ix 7.8 Transients in Static Var Control Equipment 184 Problems 185 References 186 Electromagnetic Phenomena of Importance Under Transient Conditions 188 8.1 Introduction 188 8.2. A Review of Electrostatic Induction with Some Transient Applications 189 8.3 A Review of Electromagnetic Induction and Related Topics 192 8.4 Steady-State Penetration of Magnetic Flux and Current into Conductors 200 8.5 Transient Penetration of Magnetic Flux and Current into Conductors 208 8.6 Electromagnetic Shielding 219 8.7. Implications of Electromagnetic Effects for Cryogenic Systems 228 8.8 Other Electromagnetic Manifestations of Practical Concern 229 Problems 230 References 231 Traveling Waves and Other Transients on Transmission Lines 233 9.1 Circuits with Distributed Constants 233 9.2 The Wave Equation 238 9.3 Reflection and Refraction of Traveling Waves 243 9.4 Behavior of Traveling Waves at Line Terminations 246 9.5 Lattice Diagrams 255 9.6 Attenuation and Distortion of Traveling Waves 257 9.7 Switching Operations Involving Transmission Lines 262 9.8 Multiconductor Systems and Multivelocity Waves 274 9.9 Switching Surges on an Integrated System 287 Problems 294 References 298 Principles of Transient Modeling of Power Systems and Components 300 10.1 Introduction 300 10.2 Frequency Response of Networks and Components 301 10.3 Capacitance of Windings 308 10.4 Frequency-Dependent Parameters 309il 12 13 14 CONTENTS 10.5 Circuit Reduction 312 References 321 Modeling Power Apparatus and the Behavior of Such Equipment Under Transient Conditions 322 11.1 Modeling of Transformers 322 11.2 Modeling of Generators 349 11.3. Modeling Motors 355 11.4 Model for an Overhead Transmission Line 363 11.5 Models for Cables 370 11.6 Modeling of Steel Cores 372 11.7 Miscellaneous Components 378 Problems 380 References 382 Computing Aids to the Calculation of Electrical Transients 385 12.1 Introduction 385 12.2. The Transient Network Analyzer 386 12.3 The Digital Computer 392 12.4 The Electromagnetic Transients Program (EMTP) 413 12.5 The Hybrid Computer 421 References 421 System and Component Parameter Values for Use in Transient Calculations and Means to Obtain Them by Measurement 424 13.1 Introduction 424 13.2. Transient Parameter Values for Transformers 424 13.3 Transient Parameter Values for Reactors 444 13.4 Transient Parameter Values for Generators 445 13.5 Measurement of Transient Recovery Voltages in a Power Plant 446 13.6 Transient Parameter Values for Motors 448 13.7 Transient Parameters for Transmission Lines and Cables 451 13.8 Characteristics of Bus Work 456 13.9 Capacitance of Instrument Transformers 459 Problems 460 References 461 Lightning 463 14.1 The Scope of the Lightning Problem 463 14.2. The Physical Phenomenon of Lightning 464 14.3 Interaction between Lightning and the Power System 469is 16 7 CONTENTS xi 14.4 Computation of a Specific Lightning Event 478 14.5 Induced Lightning Surges 483 14.6 Thunderstorm Tracking and Other Recent Developments 484 Problems 485 References 487 Insulation Coordination 490 15.1 Some Basic Ideas About Insulation Coordination 490 15.2 The Strength of Insulation 491 15.3. The Hierarchy of Insulation Coordination 499 15.4 Test Voltage Waveforms and Transient Ratings 501 15.5 Deterministic and Statistical Approaches to Insulation Coordination 504 Problems 510 References 511 Protection of Systems and Equipment Against Transient Overvoltages 513 16.1 Introduction 513 16.2 Protection of Transmission Lines Against Lightning 513 16.3 Lightning Shielding of Substations 517 16.4 Surge Suppressors and Lightning Arresters 517 16.5 Application of Surge Arresters 538 16.6 Surge Suppressors for Direct Current Circuits 547 16.7 Surge Capacitors and Surge Reactors 548 16.8 Surge Protection of Rotating Machines 554 16.9 Transient Voltages and Grounding Practices 556 16.10 Protection of Control Circuits 560 16.11 Surge Protection Scheme for an Industrial Drive System 566 Problems 569 References 573 Case Studies in Electrical Transients 575 17.1 Introduction 575 17.2. Misoperation of Protective and Switching Equipment During Source-Side Faults 575 17.3. Transients Associated With Bank-to-Bank Capacitor Switching 579 17.4 Voltage Escalation Due to Multiple Reignitions During Switching 583 17.5 Transient Behavior of a Transformer Coil 590xii CONTENTS 17.6 Internal Resonance in a Transformer Winding 599 17.7. An Investigation of Arrester Separation Distance 605 References 610 18 Equipment for Measuring Transients ou 18.1 Some General Observations on the Measurement of Transients 611 18.2 Frequency Response, Bandwidth, and Rise Time 613 18.3 The Cathode Ray Oscilloscope 616 18.4 Cameras for Cathode Ray Oscilloscopes 628 18.5 Magnetic Recording of Transients 628 18.6 Equipment for Measuring Transient Currents 629 18.7 Transient Voltage Measuring Equipment 641 Problems 651 References 653 19 Measuring Techniques and Surge Testing 654 19.1 Introduction 654 19.2 Minimizing Problems of Interference 654 19.3 Differential Measurements 660 19.4 Multichannel Sequence Timer 661 19.5 Low voltage Surge Testing 663 19.6 Measurement of Random Disturbances 671 19.7 Measurement of Fast Transients 678 19.8 Surge Voltage Testing 686 19.9 High Power Testing 699 19.10 Case Studies in Transient Measurements 709 Problems 723 References 724 Appendix 1 Table of Laplace Transform Pairs 727 Appendix 2 Natural Cosines 729 Appendix 3. Natural Sines 732 Appendix 4 Exponential and Hyperbolic Functions 735 Appendix § Statistical Information 737 Index 740PREFACE TO THE FIRST EDITION This book is a distillation of my experience in teaching electrical transients to successive classes of college students and practicing engineers. It also reflects fourteen years of considerable involvement with practical transient problems on electric utility and industrial power systems. Its purpose is to teach students and engineers the fundamentals of this vital subject and to equip them to recognize and solve transient problems in power networks and components. Practicality has been a paramount concern in its prepa- ration. Many of the basic notions concerning the transient behavior of electric circuits were well explored by Steinmetz and other early pioneers. What is new is the emergence and re-emergence of perennial problems in different guises with new applications and new equipment. Like successive genera- tions of cigarettes and candy bars, these problems are much the same in different wrappers. I have attempted to set out the fundamental ideas at the beginning of the book and made a consistent effort to show thereafter how one peels away the superficial differences in practical transient studies, to a point where basic principles can be applied. Where formal mathematical analysi called for, I have chosen to use the Laplace transform method. This is explained but not justified, in Chapter 2. However, there are many places in the book where solutions to problems are reached by a relatively simple process of deduction, which stresses physical insight. In such instances mathematical rigor has been subordinated to physical understanding; mathematics is often used to facilitate this understanding rather than as a substitute for it. It is my experience that the majority of students and engineers, especially those who do not have a mathematical turn of mind, proceed best by first considering the particular and then progressing to the more general. The material tends to increase in complexity as the book progresses; single-phase circuits are studied before three-phase circuits, and lumpy circuits before distributed circuits. This has one added advantage when the book is used for course text purposes. Certain chapters can be used as a basis for an undergraduate course, which could stand by itself, or lead naturally to a graduate course based on the material of other chapters. The presentation is broader in scope than most other texts on this subject, for it combines the experimental with the analytical and supplements both xiiiiv. PREFACE TO THE FIRST EDITION with many examples from actual investigations. Though basic knowledge of transients may not have advanced in recent years at the same rate as formerly, there has been a tremendous proliferation in the techniques used to study transients. The use of computers is a good example. Chapter 14 is devoted to this topic. Recent advances in instrumentation for measuring transients has been spectacular; oscilloscopes with storage tubes, sampling tubes, and traveling wave tubes are good examples. These devices and their capabilities are described in Chapter 16. Two other areas of knowledge add to the breadth of this book. In Chapter 8 I have attempted to draw from diverse places in the literature and put together as a consistent whole a collection of facts regarding certain electromagnetic phenomena that play a significant part in many transient electric disturbances. These relate to electric and magnetic coupling between circuits, more especially to the transient penetration of current and flux into conductors. These have an important bearing on such matters as pickup, shielding, attenuation or damping, and losses. The second area concerns the circuit characteristics of power system components. One may be very adept at manipulating equations, but this will be of little value unless the results can be reduced to practical terms. I have therefore included a compilation of typical characteristics of system elements, such as the capacitance and inductance of transformers, reactors, buswork, and the natural frequencies and time constants of such apparatus. I wish to acknowledge the considerable contributions that many of my colleagues and associates in the General Electrical Company have made indirectly to this book through countless discussions over the years on the subject of transients. I would like to make special mention of Dr. T. H. Lee, W. F. Skeats, and E. J. Tuohy, for most stimulating exchanges on many topics. ALLAN GREENWOOD Media, Pennsylvania March 1970PREFACE It is twenty years since I wrote the first preface; it is as relevant now as when it was written. This is not to say that nothing has changed, but that my objectives and approach remain the same. Thus, the abiding fundamentals have been left untouched in this edition, except where more recent ex- periences have changed my own perceptions and insights and where, hopefully, I have been able to improve the presentations as a consequence. During the 1950s and 1960s, engineers were slowly beginning to apply computers to the solution of power system transient problems. The past two decades have seen an enormous growth in this activity, which has been both a blessing and a curse. A blessing because it has made it possible to attack and resolve very complicated problems in a breadth and depth inconceivable in former times. A curse because some engineers have been seduced by the methods and lost sight of the physical aspects of the phenomena involved. For this reason I have continued to stress the physical while broadening and updating the computational treatment of transients in accordance with present practices. To use a computer for solving a transient problem requires some repre- sentation or model of the component or system involved. Two new chapters have been added to address the subject of modeling; models for most types of power equipment are discussed. The adequacy of models, the caution required in using them, the need for validation, and the relationship between the model and the physical entity it represents, are all stressed. A serious omission in the First Edition was any concerted treatment of insulation coordination. This has been corrected with the inclusion of Chapter 15 which is exclusively devoted to this topic. This chapter and Chapter 16 on protection, reflect the revolution that metal oxide surge arresters have caused in the power industry. More illustrative material in the way of figures and diagrams and worked examples have been included in this new edition. One entirely new chapter of case studies has been added, which demonstrates modeling and computa- tional techniques as they have been applied by practicing engineers to specific problems. References have been updated from the published work of the last twenty years, many of them from the last five years There has been a virtual explosion in the equipment for measuring electrical transients, most especially where digital techniques are involved. This has required heavy revision of the two chapters on measuring equip- xvxvi PREFACE ment and measuring test methods. This has been supplemented by the inclusion of case studies showing how the new equipment is applied. The original text has been criticized for its lack of problems that the reader could work through for himself. Accordingly, a number of such problems, well over a hundred in all, have been added at the end of most chapters. Answers are provided. It is my plan to produce a supplement of solutions in due course. I wish, again, to acknowledge the many contributions of others to this book. I would mention in particular the generations of students from whom 1 have learned so much as I have striven to teach them. Also, my sincere thanks go to Ms. Hazel Butler for typing the manuscript—a formidable task, well done. ALLAN GREENWOOD Tortola, British Virgin Islands March 19901 Fundamental Notions about Electrical Transients 1.1 INTRODUCTION An electrical transient is the outward manifestation of a sudden change in circuit conditions, as when a switch opens or closes or a fault occurs on a system. The transient period is usually very short. The fraction of their operating time that most circuits spend in the transient condition is insig- nificant compared with the time spent in the steady state. Yet these transient periods are extremely important, for it is at such times that the circuit components are subjected to the greatest stresses from excessive currents or voltages. In extreme cases damage results. This may disable a machine, shut down a plant, or black out a city, depending upon the circuit involved. For this reason a clear appreciation of events taking place during transient periods is essential for a full understanding of the behavior of electric circuits. It is unfortunate that many electrical engineers have only the haziest conception of what is happening in the circuit at such times. Indeed, some appear to view the subject as bordering on the occult. Yet transients can be understood: they can be calculated and sometimes prevented, or at least controlled, so as to be innocuous to the circuit or power system on which they appear. In this chapter we consider some basic ideas about electrical transients which will lay the ground work for their study in greater depth. 1.2, CIRCUIT PARAMETERS Examination of any electric circuit shows that it is made up of three kinds of parameter: Resistance R Inductance L Capacitance Cc All components, whether in a utility system, industrial circuit, or elsewhere, possess each of these attributes to a greater or lesser degree. Under steady-state conditions one will frequently predominate, for example, in-2. FUNDAMENTAL NOTIONS ABOUT ELECTRICAL TRANSIENTS ductance in a reactor. In the transient state, however, conditions may be very different. On occasion the distributed capacitance of the reactor winding will momentarily be its most important feature. The resistance, inductance, and capacitance of a circuit are distributed quantities; that is, each small part of the circuit possesses its share. But it is frequently found that they can be treated as “lumped” constants, concen- trated in particular branches, without seriously impairing the accuracy of calculations. We shall so treat them in much of this book. In circumstances where the technique is not suitable, as in dealing with long transmission lines, a different approach will be used. The parameters L and C are characterized by their ability to store energy, L in the magnetic field and C in the electric field of the circuit. These stored energies are functions of the instantaneous current / and voltage V, and are, respectively, 3LP and ev? In contrast, the parameter R is a dissipater of energy, the rate of dissipation being RI’ at any instant. Under steady-state conditions, the energy stored in the various induct- ances and capacitances of a direct current circuit are constant, whereas in an alternating current circuit, energy is being transferred cyclically between the Ls and Cs of the circuit as the current and voltage rise and fall at the frequency of the supply. This latter process is attended by certain losses, depending upon the resistance present. The losses will be supplied by the various sources in the system. When any sudden change occurs in a circuit, there is generally a redistri- bution of energy to meet the new conditions, and in a way, it is this that we are studying when we inquire into the nature of transients. It is very important to realize that this redistribution of energy cannot take place instantaneously for two reasons: 1. To change the magnetic energy requires a change of current. But change of current in an inductor is opposed by an emf of magnitude L dI/dt. An instantaneous change of current would therefore require an infinite voltage to bring it about. Since this is unrealizable in practice, currents in inductive circuits do not change abruptly and consequently there can be no abrupt change in the magnetic energy stored. Another way of stating this is that the magnetic flux linkage of a circuit cannot suddenly change. 2. To change the electric energy requires a change in voltage. The voltage across a capacitor is given by V= Q/C, where Q is the charge, and its rate of change isMATHEMATICAL STATEMENT OF THE PROBLEM 3 For an instantaneous change of voltage an infinite current must flow. This too in unrealizable; consequently the voltage across a capacitor cannot change abruptly nor can the energy stored in its associated electric field. The redistribution of energy following a circuit change takes a finite time, and the process during this interval, as at any other time, is governed by the principle of energy conservation, that is, the rate of supply of energy is equal to the rate of storage of energy plus the rate of energy dissipation. These three simple facts—current through an inductor cannot suddenly change; voltage across a capacitor cannot suddenly change; energy conserva- tion must be preserved at all times—are fundamental to understanding electrical transients. To fully appreciate the implications of these facts is to touch the essence of the subject. 1.3. MATHEMATICAL STATEMENT OF THE PROBLEM AND ITS PHYSICAL INTERPRETATION The statement of any circuit transient problem properly starts with the setting down of the differential equation or equations describing the be- havior of the system when excited by the particular stimulus being studied. This is usually done quite readily with the aid of Kirchhoff’s laws. Consider the very simple problem depicted in Fig. 1.1. As a consequence of closing a switch, a capacitor is charged through a resistor. To find the current, we might express the circuit equation using Kirchhoff’s first law as follows: veins 3 | rae (1.3.1) To find voltage across the capacitor, the differential equation might be written a, V=RCOTI+Y, (1.3.2) inasmuch as v ae Y ak Fig. 1.1. The RC circuit.4 FUNDAMENTAL NOTIONS ABOUT ELECTRICAL TRANSIENTS dQ _ dt Solving Eq. 1.3.2 by separating the variables, ave at V-V, RC t In(V-V,) = RE + constant or Vi=V- Ae RC (1.3.3) where A is a constant to be evaluated from the initial conditions in the circuit. If C is precharged to V,(0) before the switch is closed, setting t= 0 yields V,=V-[V-V,(0)Je "> (4.3.4) This solution is shown graphically in Fig. 1.2, which iHustrates a point made in the last section. When the capacitor is connected to the battery it does not instantaneously assume the potential of the battery but proceeds to that value through a transient, which in this instance has an exponential form. This is a simple problem, but it has all the important attributes of far more complicated problems. For this reason we will look at it in more detail. There are two recognizable parts to the solution given in Eq. 1.3.4. The first term, V, represents the final steady state when the capacitor is charged to the battery voltage. The second term is the true transient which links the initial conditions to this final steady state in a smooth, continuous manner consistent with the physical restrictions of the circuit. The form of this transient term depends essentially upon the circuit itself. The magnitude depends upon the manner in which the stored energy is disposed at time zero. This exponential will manifest itself regardless of the stimulus or drive Fig. 1.2. Capacitor voltage in the circuit of Fig. 1.1 after the switch is closed.CIRCUIT CHARACTERISTICS OR THUMBPRINTS 5 creating the disturbance. Indeed, such a circuit with no stimulus at all, left to dissipate its stored energy, would do so in this same characteristic manner: the capacitor voltage would decline exponentially if the battery was short circuited. For the circuit of Fig. 1.1, this term would be derived from the simpler equation av, RC vail +V, (13.5) which yields for a solution Vi = Vie" Mathematical texts dealing with differential equations refer to the solution obtained when the drive is set equal to zero as the complementary solution. Its physical significance is now clear; it describes the transient bridge between initial and final steady-state conditions. As stated earlier, it reflects the character of the circuit. In this instance the term ¢ "© may be described as the “thumbprint” of the RC circuit. The so-called particular solution, on the other hand, reflects the drive or stimulus creating the disturbance. When applying analytical methods to the solution of circuit problems, it is important to consider the physical interpretation of the solution reached. This will be attempted throughout the book. 1.4. CIRCUIT CHARACTERISTICS OR THUMBPRINTS It was pointed out in Section 1.3 that the single characteristic of the RC circuit which distinguishes it from other circuits is its exponential response, e—“®© to any disturbance. Since —1/RC must be dimensionless, RC has the dimensions of time; hence it is referred to as the time constant. As we have pointed out, a circuit takes a finite time to adjust from one condition to another following any disturbance. At the instant of closing or opening a switch, for example, we have certain initial conditions. Ultimately we reach a new steady state. The time constant is a measure of how rapidly this change takes place. After one time constant, 1/e of the change remains to be accomplished, or (1-1/e) has already taken place. After three time constants, conditions are within 5% of their final value. This is an appropri- ate time to look into the characteristics of other combinations of circuit elements. These elementary circuits are shown in Fig. 1.3. Close examination of these circuits reveals some startling facts. The only kind of response that is evoked when an electric circuit comprising lumped elements is disturbed takes the form of exponential functions or combina- tions thereof with real or imaginary exponents. These will sometimes combine to give sine or cosine functions. This is the case in the LC circuit.6 FUNDAMENTAL NOTIONS ABOUT ELECTRICAL TRANSIENTS e-H/RC eo RYE it. c LO 5 COR T 1.3, Thumbprints of some simple circuits. Now sines and cosines are periodic functions, which suggests the idea of a frequency. This so-called natural frequency is the thumbprint of the LC circuit. Thus we find that when such circuits are excited, no matter how, they oscillate at their natural frequencies, The LC circuit does not have a time constant because when it is stimulated it does not achieve a final steady condition but instead continues to oscillate about such a position. The period of the oscillation, which will be shown to be 24(LC)"", replaces the time constant. The RL circuit is similar to the RC circuit except that its time constant is L/R rather than RC. Through experience in handling transient problems and familiarity with solutions, the amount of formal caiculation required is diminished. It becomes possible to construct solutions in what might at first appear to be an intuitive manner. In fact, it is a consequence of consciously or uncon- sciously recognizing the thumbprints thus far discussed and applying the several other fundamental concepts outlined in the first four sections of this chapter. The only combinations of components not shown in Fig. 1.3 are the series and parallel RLC circuits. But here again such circuits react to a drive in the same manner as the simpler circuits, albeit the exponents may be more abstruse. These two circuits are given special treatment in Chapter 4. More extensive circuits are made up of combinations of the simple circuits, so that in the transient state they continue to demonstrate the same forms as their component parts. Their responses may be more complicated but they are no more complex. 1.5 THE PRINCIPLE OF SUPERPOSITION Superposition is a very important principle in many branches of physical science and a very powerful tool for solving problems, It states that in any linear system if a stimufus S$, produces a response R,, and a stimulus S, produces a response R,, then S, and S, applied simultaneously will evoke a response R, + R,. The principle is not restricted to two stimuli but is trueTHE PRINCIPLE OF SUPERPOSITION 7 for any finite number. A linear system is one in which the response is proportional to the stimulus. A simple example is Hooke’s Law, which States that the extension of a spring is proportional to the force applied to it. Thus the principle of superposition tells us that if a weight W, hung on a spring extends the spring 4,, and a weight W, causes an extension 8,, then the extension will be 6, + 6, if the weights W, and W, are attached to the spring simultaneously. The application of superposition in steady-state circuit theory is based on the linear relationship between emf and current. Thus, in a network comprising numerous branches, with say ” sources disposed around the network, the currents can be calculated in any particular branch by de- termining the sum of the currents that each source emf would drive individually. The procedure is to short circuit every source but one, leaving only the internal impedance of the remaining n—1 sources. It is then possible to obtain the current the one remaining source gives rise to in the branch of iterest. The procedure is repeated for the other sources in turn. With all n sources operating simultaneously, the current in the branch in question is the sum of the individual currents just calculated, paying due regard to their sign. Note that the principle of superposition is just as valid for the transient state as it is for the steady state, so that transients can be added to transients, or transients to steady states. We shall take advantage of this on many occasions. There are two particular applications of the principle of superposition that are of fundamenta\ importance. Earlier in this chapter it was stated that most transients are the result of switching operations. The term ‘“‘switching operation” is used in its broadest sense, meaning an event in which a new path for current is created or an existing path is eliminated. It includes the accidental application and removal of faults as well as the closing and opening operations of switches or circuit breakers. Even a lightning strike to a transmission line or adjacent structure can be considered a switching operation in that a new path for current is created. Such operations are very conveniently studied by the principle of superposition. Consider the opening of a switch in an alternating current circuit (Fig. 14a) and the subsequent interruption of the current. Usually the current is not interrupted by simply parting the switch contacts. It continues to flow through an arc that forms between the contacts; actual interruption is effected when the current comes to zero, as it does regularly twice each cycle in a.c. circuits. The current might appear as shown in Fig. 1.45. A current of this form would also be realized if, at a current zero, a current which we will designate J, were superimposed on the existing current, which we might designate /, (Fig. 1.4c). Up to instant A, /, is flowing in the circuit. After this instant, the net current flowing in the circuit is zero. Physically, we can think of this process as one in which interruption is simulated by injecting into the circuit at the contacts of the switch, a current equal in magnitude but opposite in sign to the existing current. When J, alone is8 | FUNDAMENTAL NOTIONS ABOUT ELECTRICAL TRANSIENTS ——, ad current zero (a) (6) qh Ta © Fig. 1.4. The principte of superposition applied to the opening of a switch. (a) The circuit. (6) The current. (c) Superposition of an injected current. flowing there is a certain distribution of voltage about the circuit as a consequence of the emf E. If we remove that emf and inject current J, into the circuit, another distribution of voltage would be evident. The Principle of Superposition states that when both of these stimuli, the emf E and the injected current J, are applied simultaneously, the total response will be the sum of the individual responses. Since the combination of these stimuli effectively simulates current interruption, this combined response will give the circuit’s response to the interruption and will include all the transient effects thereby evoked. The closing of a switch can be treated in a similar manner. Before closing there will be a certain voltage across the switch; it could, for example, be varying at power frequency. When the switch closes, this voltage disappears. It is as if a voltage exactly equal and opposite to that formerly existing across the switch contacts was suddenly applied at these points. By super- position, currents and voltages about the circuit after closing the switch can be obtained by adding to the currents and voltages existing with the contacts open, those stimulated by applying at the switch, the voltage appearing at the contacts before the switch was closed with its sign reversed. Recall the restriction that was placed on the application of superposition when the principle was introduced at the beginning of this section: the method can be applied only in linear circuits. There are some components in utility and industrial power systems that are nonlinear, for example, any saturable device such as an iron-cored reactor or an unloaded transformer.PROBLEMS = 9 Here the current is not directly proportional to the voltage, though this condition may be approximated over limited ranges. The application of superposition must be restricted 10 these ranges. Nonlinear resistors are used from time to time, especially as protective devices. Again, the principle of superposition should not be applied where these are located. Finally, any type of rectifier is an extremely nonlinear device since it presents almost zero impedance to the flow of current in one direction, but an almost infinite impedance to current flow in the other direction, Superposition cannot be applied indiscriminately, although it will be shown that, with care, it can be used over certain intervals, even in circuits containing rectifiers. PROBLEMS 1 Fig. 1P.1. The current in Fig. 1P.1 has already reached a steady value when S is closed. Derive an expression for the current through L after the closing of S. 1.2 If V=500 V, L =20mH and R = 30Q, calculate the voltage across the inductance 1 ms after the switch S is closed in Fig. 1P.1. 1.3 R, = 1009, R, = 10009 R, 1P.2. Initially, the capacitor C, in Fig. 1P.2 is charged to 100kV; C, is uncharged. The switch S is closed and 40 ps later the gap G sparks over. What is the current in R, and the voltage on C, immediately after sparkover?10 FUNDAMENTAL NOTIONS ABOUT ELECTRICAL TRANSIENTS 1.4 How much energy has been transferred to C, from C, at the time of gap sparkover? How much has been spent in R,? 15 s L C, = 5 uF, C, = 0.5 uF CQ C, b= 10mH V,(0) = 100 kV, V,(0) = —50 kV Fig. 1P.3. What is the maximum voltage attained by C, and the frequency of the current that flows in L, after the switch is closed in the circuit of Fig. 1P.3? 1.6 What other natural frequency could be produced by the components of Fig. 1P.3 if they were configured differently? 1.7 A capacitor C charged to voltage V is discharged into an inductor L. What is the voltage on C at the instant when its stored energy and that of the inductor are equal?2 The Laplace Transform Method of Solving Differential Equations 2.1. THE CONCEPT OF A TRANSFORM Having obtained some physical notion of what an electrical transient is, we now proceed to show how transients can be studied in a quantitative manner. Kelvin once remarked: I often say that when you can measure what you are speaking about, and express it in numbers, you know something about it; but when you cannot measure it, when you cannot express it in numbers, your knowledge is of a meager and unsatisfactory kind; it may be the beginning of knowledge, but you have scarcely in your thoughts, advanced to the stage of Science, whatever the matter may be. Today we would probably state this verity in a different way: “To under- stand something, you must be able to hang a number on it.” In this chapter we lay the groundwork for “hanging numbers” on electrical transients. The initial approach is rather formal. We use the concept of a transform. The name transform is really a contraction of a more descriptive title, a functional transformation. It implies the performing of some operation on a function to change it into a new function, frequently in a different variable. The new function is referred to as the transform of the old. Such a transformation is carried out for a purpose, in our case to simplify the solution of differential equations. There are many transform operations in everyday use in engineering which are not formally given the name, but which nevertheless are function- al transformations. Whenever we use a phasor notation to represent a sinusoidally time-varying quantity we are making a functional transforma- tion. This might also be said of the process of taking the logarithm of a number. The number is the function, its logarithm is its transform. This transformation is made to replace the processes of multiplication and division by the simpler manipulations of addition and subtraction. When such an operation has been performed, the product obtained is the trans- form of the solution, in this case its logarithm. To obtain the solution proper one must go through a reverse process, or inverse transformation, that is, W12 THE LAPLACE TRANSFORM METHOD take the antilogarithm. In most instances this last step will not be carried out formally in our analyses. Instead, we will refer to a table of functions and their transforms and extract the applicable function for our particular transform. This is akin to entering a table of logarithms to find the antilogarithm of a specific logarithm. The simple transient in the RC circuit, reviewed in Chapter 1, is a useful illustrative example, although it does not represent the type of problem regularly encountered in power systems. Practical circuits are far more complicated, so that, even after simplification for the purpose of analysis, they often retain many circuit elements in series-parallel combination. Consequently, it will require several differential or integro-differential equa- tions (one for each mesh) to describe the behavior of the circuit and each may be more complicated than Eq. 1.3.1. These equations must be solved simultaneously to evalute the variables of interest. To do this efficiently, some systematic technique must be employed. We use the Laplace trans- form method for this purpose. The Laplace transformation, when applied to terms of an ordinary differential equation, converts the equation into an algebraic equation. In so doing the variable ¢ disappears and a new variable s is introduced. The Laplace transformation has the added virtue of drawing attention to the initial conditions by providing just enough terms for these conditions to be satisfied. When operated upon in this manner the equations of the problem lose their transient aspect and appear more like equations of a steady-state problem in the new variable s. The procedure is as follows. After setting down the differential equations describing a problem, the terms are transformed one by one to obtain an algebraic equation for each of the initial differential equations. These are then solved simultaneously for the variable of interest, to give what is called the operational solution. The time function corresponding to this operational solution is then found from a table of transforms, or on rare occasions by applying the inverse Laplace transformation [1], which is a means for inverting transforms from first principles. 2.2. THE LAPLACE TRANSFORM The Laplace transform of a function F(t) is defined mathematically as follows: er@={ Fe“ dt (2.2.1) or, more precisely, lim [ Fie“ dt (2.2.2) a0 7%THE LAPLACE TRANSFORM = 13 Another symbol used for the Laplace transform of F(t) is f(s). For currents and voltages it is usual to write £I(t) = i(s) and LV(t) = v(s), reserving an uppercase letter for the function itself and a lowercase letter for its transform. We proceed with the minimum of justification for the way in which we apply and manipulate the transform, since our purpose is to use the transform as a tool rather than to study it for itself. However, there are certain questions that arise when one first applies this method. For example, are there any restrictions on F(t), or does every function have a transform? The mathematical answer to this question is that the Laplace transform can be obtained for any function of exponential order. This means any function that does not increase with t more quickly than e “ diminishes. This is another way of saying that the transform has meaning only if it is possible to perform the integrating operation described by Eq. 2.2.2. Thus we find that P is of exponential order, but e" is not, since, regardless of the value of s (as long as it is finite), as r increases, e* - e~ eventually increases indefinitely. In practical problems of circuit analysis, we are investigating the behavior of a real physical system, and to any real physical stimulus there will be a real physical response, thus in our area of interest the integral will always converge. Another question that arises is whether the Laplace transform follows the distributive law. That is, is the transform of a sum, the sum of the transforms of the parts? The answer is yes, and it can be stated thus: LUO + FO|= £F,0 + ZA (2.2.3) It will be observed that the operation of taking the transform (Eq. 2.2.2) ‘rings about a change in variable. We start with a function of ¢, F(t), and finish with a function of s, f(s). The character of s itself is relatively unrestricted. In general it can be said that s can be real or complex. It is often written s=otjo (2.2.4) Further discussion of this is left until we have developed a number of transforms. To do this, we start with some of the more common stimuli encountered in circuit problems. 1. The constant V: wve[ Ve" dt lo a vf edt (since V does not vary with f)14 THE LAPLACE TRANSFORM METHOD, " als Ss (2.2.5) 2. The ramp (typically a current ramp), a function which increases uniformly with time, J(t) = J's: £i't ={f Vte" dt J te" dt o Integrate by parts: (2.2.6) 3. The exponential e (the great prevalence of exponential functions in electric circuit theory has already been stressed): Le" a ee “dt 0 =|, eo dt lo (a-nry@ € a-slo (for s >a) (2.2.7) Note that if s RECT) (2.5.2) It is apparent from Eq. 2.5.2 that for this problem 2(s) = R + (1/Cs). Consider next the application of a constant voltage V to an RL circuit, the second problem in Section 2.3. The operational solution here is given in Eq. 2.3.10, which for our present purpose is best stated _ (9) 5 Ra Ls (2.5.3) In this expression z(s) = R + Ls. Finally, the third problem in Section 2.3 derives the current in an LC circuit, excited by applying a voltage V. Equation 2.3.23 can be rearranged in the form is -% —1 _ 2 )= > TdT (2.5.4) In this instance, 1 2)=Ls+ oe These examples indicate that z(s) is formed by writing Zs for each inductance and 1/Cs for each capacitance in the circuit. Resistors are unchanged, that is, they appear simply as R. There is a similarity between these expressions and the symbolic representation of inductive and capaci- tive reactances by jwL and 1/jwC in steady-state a.c. analysis. In fact, the latter is a special case of the former; the Laplace representation includes the steady state. Thus to solve a transient problem by use of the operational impedance, proceed as if solving for the alternating current in the branch of interest, with an alternating voltage applied. An example will make this clear. It is desired to calculate the current that would flow in the circuit shown in Fig. 2.6a when a voltage V is applied at A and B. If this was a steady-state a-c problem and V a steady-state alternating voltage, the representation of Fig. 2.6b would be used. The corresponding operationalDUHAMEL’S INTEGRAL—RESPONSE OF A CIRCUIT — 29 L job sk VOT R R R ow |. i L_. 4 BoA BA 3 1 1 Ht tt c ae 1 jot ce (a) (b) o Fig. 2.6. Comparison of a.c. symbolic representation (6) and the operational impedance (c) for an RLC circuit (a). impedance diagram for the transient problem is shown in Fig. 2.6c. Ls and 1/Cs in parallel have an impedance: LIC | s Ls +(1/Cs) — C[s? + (1/LC)} thus, 5 *W)=R+ CT GLO] is) = ——_*¥—__ (2.5.5) s s{R+ C[s*+ (/LC)] Had the stimulus been some other function, for example, 2 decaying exponential, for Ve~°’ we would have used V/(s + «) instead of the V/s used for the step function in Eq. 2.5.5. When the subject of investigation is a voltage, the operational impedance is used in the form vu(s) = i(s)z(s). The method of solution just described is best suited for problems when the circuits are initially dead, that is, for circuits that contain no stored energy. This would be the case if initially all the currents were zero and all the capacitors were discharged. Its use is not restricted to such circuits, but one must apply superposition with great care for circuits with nonzero initial conditions. For this reason, the author prefers in most instances the method described earlier of setting up the differential equations from Kirchhoff’s laws and applying the Laplace transform. 2.6 DUHAMEL’S INTEGRAL—RESPONSE OF A CIRCUIT TO AN ARBITRARY STIMULUS The material in Sections 2.5 and 1.5 (the principle of superposition) can be utilized to determine the transient response of a circuit to a stimulus of30. THE LAPLACE TRANSFORM METHOD arbitrary form. The method is formalized in Duhamel’s integral, which we will introduce shortly. Consider Fig. 2.74, which represents the waveform of a voltage surge U(t). This can be approximated by the stepped waveform shown in Fig. 2.7b. The degree of approximation will improve as the number of steps increases. Now superposition tells us that the response to a succession of stimuli can be obtained by adding the responses of the individual stimuli. In this instance, the stimuli are step functions and as Carter [2] puts it, “If the stimulus applied to a circuit consists of a succession of shocks, the response to the stimulus may be obtained by adding together the responses to these shocks.” We therefore need the response of the circuit to a step, or more precisely to a unit function, or step of unit height, often written simply 1. If we designate this response U,(¢), the response to a step of height V, that is, to V-1, will be V-U,(1). In the notation of the Laplace transform the unit function is written g4=1 (2.6.1) thus, when such a step is applied to a circuit whose operational impedance is z(s), the operational expression for the current will be WO= 56 (2.6.2) The inverse transform of the current, or what we have called u,(t) in general terms, is 1 u(Q=hQ= 2" =0 (2.6.3) uw) uy i i | | T(t#ar) , (a) (b) Fig. 2.7. Surge waveform approximated by a succession of steps.DUHAMEL'S INTEGRAL—RESPONSE OF A CIRCUIT — 31 Turning now to Fig. 2.7b, the initial value of U(s), U(0), evokes a response U(0)-u,(t). To this must be added, at appropriate intervals, the response to the other steps. Consider the one that starts at time 7; a time Ar elapses before the next step is applied. It follows, therefore, that the height of this step is U'(r)- Ar where U(r) is the value of dU/dt at the instant 7. Measured from that instant, the circuit’s response to this shock will be U'(r)u,(t— 7)Ar. Consequently the response of the circuit to the whole succession of steps up to time f is U(O)u,(t) + = U'(r)u,(t— 7) Ar (2.6.4) Where U(t) is declining, the steps are negative, but they are treated in exactly the same way. Proceeding to the limit where Az becomes indefinitely small, we find that U(t) causes a response u(t), given by u(t) = U(O)u,() +f U'(r)uy(t- 1) dr (2.6.5) This is Duhamel’s integral. Notice that in evaluating the integral, 7 is the variable; ¢ is treated as a constant. Carter [2] points out that by integrating by parts and by other elementary means, we may prove that Duhamel’s integral can be written in the following alternative ways: u(t) = U(O)u,() + [ U"(r)u,(t— 7) dr (2.6.54) ul)= uOU( + |, wu») dr (2.6.56) u(t) = U(O)u,() + [ u(t)U'(t- 1) dr (2.6.5c) u(t) = u,(0)U() + [ U(r)u((t 7) dr (2.6.5d) u(t) = 4 {f, U)u,(t~ 7) ar} (2.6.5) u()= 7 {foun ar| (2.6.5f) The choice between these different alternatives is often determined by the problem to be solved. An example will make this clear. For this example we will consider the response of the RL circuit shown in Fig. 2.8 to a stimulus32 THE LAPLACE TRANSFORM METHOD R Fig. 2.8. An RL circuit stimulated by an exponential drive U(t)= Veo" (2.6.6) We will solve this first by the regular Laplace transform method developed in the first few sections of this chapter. Subsequently, the Duhamel’s integral will be applied to show this alternative approach. The differential equation describing the circuit behavior is dy £ =Ve 2.6. IR+L 5 =Ve (2.6.7) ‘Transforming the equation gives . Vv Ri(s) + Lsi(s) — L100) = If the circuit is initally dead [i.e., 1(0) = 0], : Vv 9)" REING Fa) (2.6.8) Introducing A = R/L, this may be written 2 v = TE yETe) (2.6.9) It is apparent that Eq. 2.6.9 can be rewritten eee ( 1a ) 19)" Ta ay \sa 540 2.6.10) which contains the now familiar transform of Eq. 2.2.8 and leads to the solution 1) = Tay Mm en aty (2.6.11)PROBLEMS 33 To solve the problem by Duhamel’s integral, we must first find u,(¢), which we have already done in Section 2.3. It is apparent from Eq. 2.3.13 that 1 a u(O=L= gU-e™") (2.6.12) From Eqs. 2.6.6 and 2.6.12, it seems that the form of Duhamel’s integral given in Eq. 2.6.56 best suits this problem since the fact that u,(0)=0 simplifies the expression. Now, At wo = 2S Therefore, from Eq. 2.6. u(t) = H(t) = [ Ve” — dr = fee 7 We neiiacta which accords will solution 2.6.11. We have been thinking in terms of a voltage stimulus, but it should be clearly understood that Duhamel’s integral can be applied equally well if U(O) is current stimulus. In this case u,(t) will be a voltage and will be obtained by multiplying the current by the impedance: uy(s) = 4-268) or ue et PROBLEMS 2.1 Using the method of partial fractions, evaluate the following Laplace transforms (obtain the time function): a ul * +5524 1s? +4a° * s(s? +a’) 2s? +135 +1 S +78? + 6534 THE LAPLACE TRANSFORM METHOD 2.2. The transform of a certain voltage is given by: 2.3 24 2.5 2.6 1.9x 10" s°+2.1x10°s +2 10" Evaluate the transform and sketch its form with reasonable accuracy. How much energy will be dissipated when the switch in the circuit in Fig. 2P.1 is closed. C, = 60 uF C, = 40 uF R=59 Fig. 2P.1. The capacitor C, in Fig. 2P.1 has an initial charge of 1.0C; C, is discharged. Calculate the following: a. The peak current b. The current 200 ws after the switch closes c. The ultimate energy stored in C, d. The ultimate voltage on C, If the resistor in Problem 2.3 is replaced by an inductor with the same 60 Hz reactance, calculate the following, once the switch is closed: a. The instantaneous current b. The peak current c. The energy stored in the inductance 1 ms after the switch is closed d. The energy stored in C, at the same instant. Show that if one capacitor is discharged into another through a resistor, the energy dissipated in the resistor is independent of the value of the resistor. Each phase of a 3-phase capacitor bank is rated 60MVA at 13.8/ V3 kV. A second bank has a rating of 30 MVA at 13.8/V3 kV. The two are to be paralled by momentarily connecting them through a 1000 stainless steel resistor (one for each phase), which will be subsequently shorted out. You are to design these resistors (determine the length and cross-sectional area of the wire to be used) if the temperature rise of a resistor is not to exceed 200°C, when the switching operation is made at a time when one capacitor is at positive peak voltage and the other at negative peak voltage.2.7 2.8 PROBLEMS 35 The characteristics of stainless steel are: density = 7.9 g/cm’; specific heat = 0.5J/g per °C; resistivity = 72 Qcm. Assume that no heat is lost to the surroundings during the switching operation. What will be the weight of the resisitor? What will be the peak current during the switching operation? Field coil: L = 2H, R= 3.62 R,= 109 Fig. 2P.2. Figure 2P.2 shows the field coil of a machine. It is excited by closing switch S, onto an 800 V d.c. bus. Determine the energy stored in the coil, and the energy already dissipated in it, 1s after S, is closed. When the coil current has attained a steady value, S, is opened and S, is closed simultaneously. What will be the voltage across 5, 0.1s later? How much energy will eventually be dissipated in R,? We are often required to design test circuits which will generate surges of specific waveform. These are then used to apply surges to pieces of power equipment (transformers, generators, reactors, etc.) we wish to test. Sometimes we wish to simulate the effect of a lightning surge, sometimes a switching surge. Fig. 2P.3. Figure 2P.3 shows a basic form of impulse generator. When C, has been charged and the gap G is caused to spark over, an impulse voltage is generated at the output terminals A and B. Without solving the equation of the circuit, compute a good estimate of the following when the precharge voltage is 500kV and the gap discharges.36 THE LAPLACE TRANSFORM METHOD a. The maximum current in R, b. The maximum voltage across C, c. The time when this voltage (b) is reached d. The output voltage after 0.5 ps e. The output voltage after 50 ys 2.9 V=250V v L R=089 L=0.4H Fig. 2P.4. R and L in Fig. 2P.4 represent the resistance and inductance of the field winding of a machine. The switch S has been closed and a steady direct current is flowing from the source V. When S is opened, an arc is established between its contacts which develops a voltage of 400 V, opposing the flow of current. Plot the current after S opens. REFERENCES 1. $. Goldman, Laplace Transform Theory and Electrical Transients, Dover Publica- tions, New York (1966). 2. G. W. Carter, The Simple Calculation of Electrical Transients, Cambridge Uni- versity Press, New York (1944).3 Simple Switching Transients 3.1 INTRODUCTION It was pointed out in Chapter 1 that a transient is initated whenever there is a sudden change of circuit conditions; this most frequently occurs when a switching operation takes place. All the samples studied thus far have been concerned with switching transients, but they have been confined to circuits with d.c. drives. In this chapter we consider a.c. circuits, restricting our- selves to two problems: (1) the closing of a switch or circuit breaker to energize a load and (2) the opening of a breaker to clear a fault. This provides an opportunity to discuss some of the practical details of switching transients. A prime concern throughout this text is the emphasis on the physical aspects of what is occurring in the circuit. Mathematics is used as an adjunct to this end, not as a substitute for it. The Laplace transform method developed in Chapter 2 has helped to simplify and systematize the mathe- matics. However, where there are opportunities to simplify the mathematics further, we shall take them. Ultimately, we will find that we can circumvent the mathematics entirely in many of the problems we are about to discuss, proceeding directly to the solution by deduction. 3.2. THE CIRCUIT CLOSING TRANSIENT Example 1: Energizing an RL Circuit. The circuit involved in this example is that of Fig. 3.1. It has been reduced to its barest essentials in the interest of initial simplicity, The load is represented by a series combination of resistance and inductance, which has a steady-state power factor given by R "@roLy an R ez) The source is assumed to have negligible impedance compared with the load. The source voltage is V, indicating a phasor varying at the supply frequency w. When the switch S is closed, the equation expressing the current is 3738 © SIMPLE SWITCHING TRANSIENTS Ss R v L Fig. 3.1. An RL circuit with a sine wave drive. dl . RI+ LF =V=V,,sin (or + 8) (3.2.2) The inclusion of the arbitrary phase angle @ permits closing of the switch at any instant in the voltage cycle Before attempting to solve Eq. 3.2.2, let us consider for a moment what we have already discovered about this circuit. It is clear that in due course the current will attain a steady-state value of V/Z, and that it will lag in phase the voltage by an angle ¢ defined by Eq. 3.2.1. However, it is equally clear that except perhaps for some special circumstance, the current cannot achieve this value instantaneously, because the circuit inductance demands that the current start at zero. We would suppose, therefore, that there is some transient that leads the current to its steady-state value in a smooth, continuous way and that since this is an RL circuit, the exponential « *"’" will play an important part in the solution. These observations say a great deal about the solution of Eq. 3.2.2, although as yet we have made no attempt to solve it. We now proceed to this task. Equation 3.2.2 can be rewritten dl RI+L] V,,(sin wt cos @ + cos wt sin 6) Transforming both sides, @ COs O Ss ne) + Ri(s) + Lsi(s) — LI(0) = ¥A(4 soit Fre? (3.2.3) Remember that sin @ and cos @ are constants once the value of @ has been assigned. In this circuit (0) =0, so the operational solution for current is gy= Ya 1 (aoe , ssin0 ) 5) = Y= _1___ (woos 3.2.4 = TT FRI) NS +0? Fa? G24) This can be written more consisely as follows: _ B: i(s)= a (3.2.5) Grayleto) * Grae to)THE CIRCUIT CLOSING TRANSIENT — 39. where Aa‘ 0 B= “sing = “fr #008 8, =Frsin, a=F These are new transforms, but they can be reduced readily by the method of partial fractions outlined in Section 2.4. Simple manipulation reveals that 1 1 oe (stalst+o) (ato) \sta ste +o” (3.2.6) Therefore _ Cr (stas? +o") (a? +07) (e« —cos wt + © sin ws) (3.2.7) The other terms, Bs/(s + )(s” + w”) in Eq. 3.2.5 can be evaluated by the same method. Instead let us turn around Eq. 2.2.12 and obtain the inverse transform of this second term by differentiating Eq. 3.2.7: 1 1 (a° +’) (s+ a)(s’ + w*) xX (-ae" + wsin wt + acoswt+1—1+0) (3.2.8) Equation 3.2.4 can now be evaluated with the aid of Eqs. 3.2.7 and 3.2.8: V, 1)= Tate) [o cos ole — cos wt + = sin wt) + sin 0(@ cos wt + w sin wt — ae") V, = Tea? ra’) Wo 0080 - asin ae" —(@ cos @— a sin 6) cos wt + (a cos 6 + w sin @) sin wt] (3.2.9) V2 Now from Eq. 3.2.1, tan g = wL/R = w/a, so that sin g = w/(a’ + 0) and cos g = a/(a* + w”)'””, Equation 3.2.9 simplifies further: V, t= Teta [-sin (0 — g)e “+ sin (wt + 0 - ¢)] y, = a TET Miia (ot + Oo) ~ sin (0 ee] (3.2.10)40 SIMPLE SWITCHING TRANSIENTS. This is precisely the form we predicted before embarking on the analysis. The first term is the steady-state final value. Its amplitude is V,,/|Z| and it indeed has a phase angle — ¢ with respect to the voltage. The second term is the transient. It involves, as expected, the exponential « *“; moreover, at t=0, it is equal and opposite to the steady-state term, thus assuring that the current start from zero. Equation 3.2.10 is depicted graphically in Fig. 3.2. In the very special case where the switch closes at the instant when 6 = ¢, the transient term will be zero and the current wave will be symmetrical. On the other hand, if the switch closes when 6 — y = + 77/2, the transient term attains its maximum amplitude and the first peak of the resulting composite current wave will approach twice the peak amplitude of the steady-state sinusoidal component. This has some significant practical implications for circuit breakers. An important function of a circuit breaker is to close circuits as well as open them and on occasions a breaker will close into a short circuit. The R and L in Fig. 3.1 represent the resistance and induction of the source in this context; closing the switch is equivalent to the occurrence of a fault. The closing angle @ is arbitrary, so it is possible to hit the condition where the current is fully offset for which the peak of the first loop approaches a maximum value of twice the peak short circuit current. In a three-phase circuit, the closing angles in the different phases are 120° apart if the poles close in mechanical synchronism, so there is a good chance that this condition, or something close to it, will be met on some phase at every closing operation. The circuit breaker must be designed electrically, mech- anically, and thermally to withstand this duty and perform the subsequent operations (such as opening and reclosing) without its effectiveness being impaired. Electromagnetic forces set up by such currents could cause buckling of parts in an inadequately designed circuit breaker. It should be remembered that these forces are proportional to the square of the current. Thus, if the current is doubled, the forces are increased fourfold. ‘Steady-state term @-e—4 — “Ye Gee ctr SPO Resultant current I(t) Fig. 3.2. Asymmetrical alternating current from Eq. 3.2.10.THE CIRCUIT CLOSING TRANSIENT 41 Similar asymmetrical currents can flow through the contacts of a closed breaker if a short circuit occurs on the sysem it is protecting. These so-called high “momentary currents” will encourage the contacts to weld at the points where they touch, due to the energy dissipated there in contact resistance. In configurations where the contacts are butted together this condition may be made worse by the contacts “popping,” that is, being momentarily driven apart, with the establishment of an arc, by the electromagnetic forces of the currents converging on and diverging from the points of contact. Again, any weld so formed must not impair the efficiency of the circuit breaker. Its mechanism must be able to break the weld and subsequently rupture the fault current if called upon to do so. Circuit breakers frequently have to interrupt asymmetrical currents, for a fault can occur at any point in the cycle. Some may not be able to do this at the current zero following a major loop of current. The arc will reignite and interruption will be effected after the minor loop. However, when we consider recovery voltages it will be apparent that current amplitude is not the only influential factor in determining whether a breaker will successfully interrupt. The rate of decay of the transient component of an asymmetrical current depends upon the time constant of the circuit. It is often referred to as the d.c. decrement of the circuit. It varies with the type of circuit and is usually expressed in terms of the X/R ratio rather than the L/R time constant. Typical X/R ratios for different circuits are given in Chapter 13. If we know the time from the initiation of the fault to the parting of the circuit breaker contacts and the available fault current on the system, it is possible to determine the actual current the breaker has to interrupt. Clearly in a circuit with a high decrement, the duty will be less severe than in one where the time constant is long. It is not usual for the opening of a breaker to be deliberately delayed in order that the current may become more symmetrical. Such considerations as limiting of damage at the fault, thermal overloading of lines and equipment, and system stability make it mandatory to remove the fault as soon as possible. Figure 3.3 gives some oscilloscopic records of closing transients showing the effect of varying the instant in the cycle at which current is initiated. Example 2: Capacitor Inrush Currents. Shunt capacitors, connected be- tween line and neutral or line and ground, are common sights on power systems. They may be there to correct a lagging power factor, or in some cases, to provide voltage support for the system. In some applications they are switched in and out quite frequently as the system load varies and the system voltage fluctuates. It will be shown that such switching operations are nontrivial and should be carefully considered when designing capacitor banks and their associated switching equipment [1,2]. In this example we focus on the closing operation or energization of the capacitor bank. Quite often the capacitor bank is divided into a number of parallel42 SIMPLE SWITCHING TRANSIENTS pene ee 419K) Fig. 3.3. Examples of transient currents showing various degrees of symmetry. (2) Symmetrical. (b) Partially asymmetrical. (c) Fully asymmetrical.THE CIRCUIT CLOSING TRANSIENT 43 o--—————— Feeders Source Capacitor Bus q x. I Fig. 3.4. Switching capacitors on a substation bus sections which can be switched independently. In Fig. 3.4 there are two sections with their switches S, and S,. The source inductance is represented by L, while L, represents the inductance of the local circuit comprising the two capacitors and their bus and ground connections. Resistance is assumed to be negligible. Consider first the closing of switch S,. It was shown in Section 2.3 (Eq. 2.3.24) that the current that flows when an LC circuit is energized is given by 1) = YO sin ogt (3.2.11) 0 where Z,=(L/C)'? and V(0) is the instantaneous voltage driving the current. If the capacitor is discharged V(0) will be the instantaneous supply voltage. If the capacitor has an initial charge giving it a voltage V;(0), this must be added to the supply voltage paying due regard to sign. Put another way, V(0) is the instantaneous voltage across the switch at the moment of closing. For the purpose of calculating the transient inrush current, it is common practice to treat this as a constant voltage since the frequency of the transient current is usually much higher than the power frequency, so that the power frequency voltage changes very little during the period of interest. Relating Eq. 3.2.11 specifically to Fig. 3.4 we have cy 19 = vo $) sin ER (3.2.12) Consider a 34.5kV system, solidly grounded, with an available short circuit current at the bus of 25kA rms, symmetrical. Assume C, = 18 MVA bank and C,=10MVA bank: for C,, V3V/=1.8x 10", whence normal44 SIMPLE SWITCHING TRANSIENTS. Fig. 3.5. Schematic diagram of capac- oo _| itor banks. 60 Hz current 2 eels x07 ~ V3x 34.5 x 10° 34,500 = SP = 66.130 V3 x 301.2 I =301.2A thus, C,=40.1 uF Source reactance = 34,5/V3 x 25 = 0.797 0, so L = 0.0021 H (=2100 #H) Peak inrush current if the switch closes at voltage peak (34.5V2/V3)(40.1/2100)'/? = 3.893 kA This is more than 9 X peak 60 Hz current. Frequency = 1/27(LC,)'” = 548 Hz (justifies assumption of constant source voltage). Consider next the closing of switch S,. As a first approximation we will neglect current from the source. This simplifies the circuit, leaving C,, Cy and L, series-connected. L, depends entirely on the geometry or physical dimensions of the capacitor bank. A typical arrangement might be as shown in Fig. 3.5. The inductance of buswork depends on the size of the conductor. For aluminum angles and channels often used in capacitor banks, a figure of 0.2 wH/ft is reasonable. For thinner conductors it could be 0.4 wH/ft. Using 6.3 wH/ft gives L, =0.3 x 64 = 19.2 wH E OG _ C,= 22.28 uF 80 CG = M32 uFTHE CIRCUIT CLOSING TRANSIENT — 45 The surge impedance of the circuit = Z,, = (19.2/14.32)"? = 1.1580. In- rush current from C, to C,, assuming C, discharged and C, charged to peak system voltage = Vpeax/Zo, = (34-5V2) /(V3 X 1.158) = 22.33 kA! 10° er = 9.60 KH: 2m(19.2 x 14.32)? 2 Frequency of inrush current = This justifies neglecting current from the source. If C, has recently been disconnected and is being re-energized, it is entirely possible that it will retain charge from its previous energization. In the extreme case, this trapped charge could give C, a voltage almost equal in magnitude and opposite in sign to C,. The peak inrush current could then approach 48 kA. Such excessive high frequency currents can create problems for a number of reasons. First, they produce severe mechanical stresses. Second, they can induce undesirable transients in neighboring circuits; low power relay and control circuits being particularly vulnerable. This problem is discussed in Chapter 8. A variety of methods have been considered and used for reducing transient inrush currents on capacitor bank switching. The surge impedance can be increased by the intentional addition of inductance in the form of a reactor. Resistance can be added to damp the oscillation (see Chapter 4), but unless this is shorted out following the operation, the steady-state losses are greatly increased. Shorting out requires a second switch which is more elaborate and costiy. A critique of different solutions has been published by O'Leary and Harner [3]. Another approach is to use synchronous switching which means closing the switching device at a selected point in the cycle. This is done, ideally, when there is no voltage between the contacts. It requires careful sensing, precise and consistent mechanical operation of the switch and independent operation of the poles of the three phases. Moreover, there is always the problem of prestrike, the tendency for the intercontact gap to breakdown and establish current before the contacts physically touch, because of the voltage stress between them. For these reasons the method has not been widely used. The source was ignored in the bank-to-bank switching operation ex- amined above. Its effect is felt, however, in due course. The first rapid transient brings about an exchange of charge between the capacitors being switched; they are brought to a common voltage since losses damp out the transient in practical installations. The common voltage is different from the supply voltage, so a second transient follows during which the two capacitor banks are restored to supply potential. The initial inrush transient and the restoring transient can usually be treated separately, since the first is usually completed before the second has really started.46 SIMPLE SWITCHING TRANSIENTS The common voltage following inrush can be determined by considering charge conservation. The procedure will be illustrated with the example of Fig. 3.4. It will be assumed that C, had a residual voltage of —8 kV at the time of switching and that the supply voltage was at a positive peak. Since charge is conserved when the capacitors are connected together, E,V,(0) + C.V4(0) = (C, + CV) (3.2.13) where V(~) is the subsequent common voltage. Thus, V(~) = VO) + CoV (0) _ 40.1 x 28.17 = 22.288 J C+G ~ 40.1 + 22.28 = 15.25kV The instantaneous supply voltage is 28.17 kV (peak system voltage) so that, neglecting damping, the two capacitors will swing as far above this value, as they started below, i.e. from 15.25kV to 41.09kV. The frequency of this restoring oscillation will be a St a9. LULL [ The current involved is given by C+G)\'? L sin ot wv, - veon( = (28.17 — 15.25)( 238) sin wt = 2.227 sin wtkA (3.2.14) The whole transient disturbance is depicted in Fig. 3.6. Fig. 3.6. Voltage oscillations during bank-to-bank capacitor switching.THE REMOVAL OF A SHORT CIRCUIT 47 3.3. THE RECOVERY TRANSIENT INITIATED BY THE REMOVAL OF A SHORT CIRCUIT The simplest circuit that can be chosen to illustrate this phenomenon is that shown in Fig. 3.7. It contains only what is necessary for the purpose. It is assumed that the load being fed through the circuit breaker is suddenly isolated by the occurrence of a fault and that the fault is a dead short circuit. The L is all the inductance limiting the current to the point of fault, and C is the natural capacitance of the circuit adjacent to the circuit breaker. Such capacitance always exists. It comprises capacitance to ground through bushings, current transformers, and so forth, and perhaps the capacitance of an adjacent transformer, as well as capacitance across the breaker contacts that will be grounded on one side by the fault. Circuit resistance and other forms of loss will be neglected at this time. When a fault occurs on a system, a substantial fault current usually flows. ‘The parting of the circuit breaker contacts does not in itself interrupt the current, for an arc will be established between the parting contacts, through which the current will continue to flow. Successful interruption depends upon controlling and finally extinguishing the arc. Different types of circuit breaker use a variety of different agencies to bring this about. Fortunately, in alternating current circuits at least, the current passes through zero twice each cycle. It is at one of these instants that interruption is actually effected. In the circuit of Fig. 3.7 the current is assumed to be symmetrical and will be completely reactive since it is limited entirely by inductance. This means that at the instant of current zero, the circuit voltage will be at a maximum value, but the voltage at the switch contacts, and therefore across the capacitor C, will be the arc voltage. The relative importance of the arc voltage varies. In high-voltage circuits it is usually only a small percentage of the system voltage. In low-voltage circuits it may be much more significant. For the first examination of the ensuing events, it will be assumed that the arc voltage is negligible. In this analysis, time will be measured from the instant of interruption when the fault current has just come to zero. Since the source voltage is a ic Circuit breaker a. ——S 6 ——_-x Source V= ck Vn COs wt ale Fault x Fig. 3.7. Equivalent circuit for studying the transient recovery voltage when a circuit breaker clears a fault.48 SIMPLE SWITCHING TRANSIENTS. sinusoidally varying quantity and is at its peak at the moment, it is expressed as V,, cos wt. The circuit equation is therefore Lo + V=V, cos wt (3.3.1) There are two unknowns, / and V¢, so another relationship between them is required. This is ay, iG oe (3.3.2) for after the switch has cleared, the only path for current is into the capacitor. The combination of these equations gives (3.3.3) Before solving this equation, let us consider, as we did with the closing transient in Section 3.2, what kind of a solution we might expect from the physical conditions in the circuit. Ultimately, after the circuit breaker has completely cleared the fault, we would expect to see across its contacts the voltage of the supply. But at the moment of clearing (¢ = 0), the voltage is the previous arc voltage, which we have chosen to neglect. The voltage cannot change discontinuously because capacitor C must be charged. We would therefore expect that following current zero, a transient would be initiated whereby C is charged from the supply through the inductance L. Our experience with LC circuits in Chapter 2 leads us to expect that this will cause an oscillation at the natural frequency of the circuit. The analysis is very similar to that used in Section 2.3. Let 1/LC = a5, and transform Eq. 3.3 5*¥¢(s) — SVe(0) ~ Ve(0) + waves) = WV, > m Fa gt or a 5 s V_(0) As) = @2V,, ——z-—y + Ve(0) + vel) = 0m ET DAGEF at) Oras areas (3.3.4) If we neglect arc voltage the second term on the right is zero. The third term is also zero because, from Eq. 3.3.2,THE REMOVAL OF A SHORT CIRCUIT 49 Therefore only the first term remains. Now ( : : ) oS to S$ +03 Therefore and Vet) = 3 V, (cos wf — cos wot) (3.3.5) 2 5 -o Equation 3.3.5 is the voltage appearing across the switch contacts after current zero. It is the classical transient recovery voltage (TRV) described by Park and Skeats [4]. Almost invariably w, > w, so that @5/(w3 — w?) 1. Thus to a very close approximation, V.(t) = V,,(COs wt — cos wyt) (3.3.6) Indeed, it often happens that over the period of interest (the time for which the natural frequency oscillation persists), there is little change in the power frequency term. In this case Eq. 3.3.6 can be further reduced to give Velt) = Vp{1 — cos wot] (3.3.7) This is very evident from Figure 3.9 which shows an oscillogram of the current interruption period. In this case the TRV oscillation (lower trace) lasts for about 600 ys. The decline of the current to zero can be seen in the upper trace. It will be noted that the TRV begins with a small excursion of the opposite polarity to the instantaneous system voltage. This indicates some current chopping, a subject discussed in Section 5.2. Events before and after current zero are depicted in Fig. 3.8. This diagram shows the accuracy of the predictions we made before starting the analysis. It was pointed out then that there would be losses in the system that would damp out any disturbance created by the opening operation. This damping is not included in Eq. 3.3.6 but has been introduced into Fig. 3.8 and is evident in Fig. 3.9. Observe how the voltage approaches twice the peak system voltage when the two components interfere constructively. Up to the instant of current zero, the capacitance in Fig. 3.7 has been short circuited by the fault. With the successful operation of the circuit breaker this constraint is removed; the source will therefore attempt to charge the capacitance to its own potential. The situation is quite similar to50 SIMPLE SWITCHING TRANSIENTS Transient recovery voltage, Vert) System voltage 77 Fault current 3.8. Transient recovery voltage across the circuit breaker in Fig. 3.7 following interruption of the fault current that depicted as curve (b) in Fig. 2.4 and we see the inertial effect of the circuit inductance causes the overswing of voltage we observed there. If the natural frequency «, is high (if L or C or both are very small), the voltage across the switch contacts will rise very quickly; this is the case in Fig. 3.9. If this rate of application of the recovery voltage should exceed the rate of buildup of dielectric strength in the medium between the contacts, the breaker will be unable to hold off the voltage and a so-called reignition will occur. This usually results in the switch carrying fault current for at least another half cycle. It is for this reason that in British literature the recovery transient is often referred to as the restriking voltage. (Perhaps the two different names for the same phenomenon, recovery voltage and restriking Fig. 3.9. Circuit breaker transient recovery voltage (lower trace) obtained ex- perimentally with corresponding current (upper trace).THE REMOVAL OF A SHORT CIRCUIT 51. voltage, reflect the basic optimism or pessimism of American and British switchgear engineers!) It is apparent that the rate of rise of the recovery voltage, r.r.r.v., is an important factor is switchgear application. It gives a measure of circuit severity from a switchgear point of view. As already mentioned circuits that lead to the highest r.r.r.v.s are those with high natural frequencies. An air-cored reactor is a good example. Consider such a choke with an inductance of 1mH and an effective capacitance of 400 pF, its natural frequency will be 1 10° SERIO = 3,95 ~ 250 kHz fo= 3.99 2n(10 This represents a period of 4 zs. On a 13.8-KV circuit, neglecting damping, the voltage will swing to twice the system peak in half a period. The mean rate of rise of voltage during this time is therefore 2x 13.8V2 V3x2 > 11.3 kV.ps This is beyond the clearing capability of many circuit breakers. Another example of a very fast recovery transient is that initiated on the clearing of a “kilometric fault”. Thi iscussed in Chapter 9. It was shown in Section 3.2 that when a switch is closed at random it is likely that the current will lack symmetry. It can have any degree of asymmetry, depending upon the instant in the voltage cycle at which the switch is closed. Similarly, a fault current can have any degree of asymmetry depending upon the time in the cycle at which the fault occurs. The circuit breaker will again interrupt at current zero and the recovery voltage will oscillate about the instantaneous value of the supply voltage. But with an asymmetrical fault current the voltage will no longer be at its peak. The transient recovery voltage is therefore not as high. This is illustrated in Fig 3.10. In the preceding analysis, specifically in Eq. 3.3.4, the arc voltage in the circuit breaker was assumed negligible, although it was pointed out that sometimes this is not the case. For example, consider the arc voltage term in Eq. 3.3.4. Its inverse transform gives a cosine at the natural frequency , B'V(0) Sy = Ve(0) £08 at fy Now V,(0) is the value of the arc voltage at the time of current zero (t = 0), since the capacitor always has the arc voltage impressed across it as long as the arc persists. This adds to a similar term in the recovery voltage tending to increase this switching transient. The effect is offset by a second effect of52 SIMPLE SWITCHING TRANSIENTS Transient Tecovery voltage ‘Asymmetrical fault current Fig. 3.10. Voltage transient following the interruption of an asymmetrical current. the arc voltage, which is to oppose the current flow and thereby change the phase of the current, bringing it more into phase with the supply voltage. Thus, when the switch clears, the supply voltage is not at its peak. These two effects are illustrated in Fig. 3.11, which should be compared with Fig. 3.8, where the arc voltage has been ignored. 3.4. DOUBLE-FREQUENCY TRANSIENTS The simplest form of the double-frequency transient is that initiated by opening the circuit breaker in the circuit shown in Fig. 3.12. In this diagram L, and C, are the inductance and stray capacitance on the source side of the breaker. Here L, and C, might represent aa inductive load and its stray capacitance (e.g., an unloaded transformer). When the switch operates in such a circuit it completely divorces the load from the supply. Thereafter the two halves of the circuit behave independently. It is possible to deduce what happens following such a switching operation without resorting to any Transient recovery voltage System voltage Fig. 3.11, The effect of arc voltage on the recovery transient after switching.DOUBLE-FREQEUNCY TRANSIENTS — 53. Ly rc Source C a oy zy Fig. 3.12. Circuit with two natural frequencies. mathematical analysis. Before the switch opening, the 60 Hz voltage will divide in proportion to the inductances, that is, to a close approximation the voltage of the capacitors will be It is likely that L, > L,, otherwise the regulation would be very poor, so that C, and C, are charged to something a little less than the instantaneous system voltage. When the current passes through zero, this voltage will be at its peak. Following current interruption C, will discharge through L, with a natural frequency given by al mL" oe f Meanwhile C,, now free to take up the source potential, will oscillate about that value until the losses of the system damp out the disturbance. The frequency of this oscillation will be iT * ae. GN™ es) Source side and load side transients are depicted in Fig. 3.13a and 3.13b. The recovery voltage across the circuit breaker contacts will be the differ- ence between these two, as shown in Fig. 3.13c. There are many other two-frequency circuits that occur in practice. One that is encountered quite often is drawn in Fig. 3.14, which also shows one situation in which it might arise. This shows a circuit breaker clearing a short circuit on the secondary side of a transformer. In this case L, represents the inductance up to the transformer, L, the leakage inductance of the trans- former, while C, and C, are the inherent system capacitances on either side of the transformer. We recognize this as a double-frequency circuit by the two LC loops. However, it is by no means clear what the natural frequencies are because54 SIMPLE SWITCHING TRANSIENTS (a) (o) Fig. 3.13. Double frequency recovery transient. (a) Source side transient. (b) Load side transient. (c) Recovery voltage across the switch. Ly la tax TTT s b AY 1 a ve v “) G *) C2) x Transformer Fault 3.14. Example of a two-frequency circuit,DOUBLE-FREQUENCY TRANSIENTS — 55 the loops are mutually coupled. Still more obscure is the relative amplitude of the two component frequencies. To determine the recovery transient we must analyze the circuit. Contrary to what one might suppose, the algebra rapidly becomes quite complicated. We assume that during the period of interest the change in the source voltage is negligible. It can then be represented by a constant voltage V. Because the reactance of C, at 60Hz is very much higher than the corresponding reactances of L, and Ly, it is safe to assume that the initial voltage distribution is determined by the Ls or, more specifically, that the initial voltage on C, is given by L. Z,+L. V.,(0) = v (3.4.3) The following equations can be written: dl, 1 AVR Ly GA Ve lO) + | (1, - 1) dt (3.4.4) 1 Yer & i, I, dt (3.4.5) dl, dl, Vo=V~ Ly Gtk (3.4.6) 2 dt When transformed, these equations read as follows: Ynys) — Ey) =O + A tio) 100 or : V.(0) ~is)(L Lys + a) + wo = + L110) ~ Y (3.4.7) ve(8) = a) (3.4.8) _(s) = ¥ — Lysi,(s) + Ly1,(0) — Lysi(s) + LyI,(0) (3.4.9) Time is measured from the instant when the switch clears at current zero, so that /,(0) = /,(0) = 0. From Eqs. 3.4.8 and 3.4.9, v - (Br +1 W)= 7a rag Joos) (3.4.10) Substituting Eqs. 3.4.10 and 3.4.8 in Eq. 3.4.7,56 SIMPLE SWITCHING TRANSIENTS. LCs" +1) pe K 1 ) CQ _ Ve(0)-V (AF ono - fale + A) +B eo= (3.4.11) Multiplying through Eq. 3.4.11 by s7/L,C, and rearranging, [ na) pene) T,C,/" L,G,1,¢, I’o* _ Me,(0) Vv ~ v — c)* none] 7,6) * LGLe) a 1 “(a rane @ Tene) (G.4.12) The expression on the left-hand side of Eq. 3.4.12 is a quadratic in s*. The equation can therefore be rewritten: (s+ 03)(s? + 03)v¢(6) = av(4 + Bs) or Bs | (8° + wi )(s* + @3) Pes) = av za eee G3) in which ,, w,, A, and B are constants, dependent upon the circuit parameters L,, C,, L,, and C,. Inverse transforms for the two terms can be obtained by taking partial fractions or by looking up the inverse transform in the table in Appendix 1. From there we find the solution to be vo=av[a+ 2s 10, (0; — + a 3 (cos wyt — cos 0] @;—@2 1 [ 1- ofB 1-03B | ‘| =z ST) | COS Wot @(@;, ~ ©) emeena) @3(@; — @3) ~ay{ Joos a @103 (3.4.14)PROBLEMS 57 It is clearly possible to write Eq. 3.4.14 in terms of the circuit constants L,, C,, C,, and C,, but the expressions are very lengthy because of the complicated relationships between these quantities and w, and w,. At this point (i.e., Eq. 3.4.14) in a practical problem, one is well advised to substitute numbers before proceeding. This exercise shows, if nothing else, how quickly the algebra and arith- metic burgeon with a modest increase in the complexity of the problem. For this reason we explore in later chapters methods whereby some of this labor can be avoided, most notably by the use of a digital computer. PROBLEMS 3.1 3.2 3.3 Viewed from the point of fault, a three-phase system can be repre- sented by an ideal three-phase, 13.8 kV (rms) generator, with a series impedance of (0.02 + j0.43) 0/phase. The system is solidly grounded. A fault-to-ground occurs on one phase when the instantaneous vol- tage is 3.5kV and declining. Calculate the approximate value of the first and second peaks of the fault current. Calculate approximately the maximum force/meter on the phase and ground buses at a point where they run parallel, 20cm apart. A c oY, a Fig. 3P.1. Without formal analysis involving differential equations, determine the voltage V; across the capacitor in Fig. 3P.1 when a. A step of voltage, V, is applied at the terminals A and B from an infinite bus b. A ramp of current, /’r, is injected from a current source into the terminals A and B Figure 3P.2 shows two capacitor banks, C, and C;, in a substation. C, is energized, but C, is discharged. The three-phase, 60 Hz ratings of the banks are: C,, 5 MVA; C, 3 MVA, on a 13.8kV base. The source has a short circuit rating of 20kA rms at 13.8 kV. The inductance of the loop between C, and C,, represented by L, is 30 #H.58 SIMPLE SWITCHING TRANSIENTS Source Capacitor banks Fig. 3P.2. Calculate the peak transient voltage that will appear on C, and the peak transient current that will flow in L,, if the switch S is closed at the peak of the voltage cycle. Point out any assumptions you make. 3.4 bm o ac Fig. 3P.3. = Figure 3P.3 represents one phase of a three-phase, 69 kV circuit, containing a source and a capacitor bank C, which is rated at 15 MVA/phase. L = 60 pH. Calculate: a. The peak voltage that can be attained by C when the switch is closed and C has an initial voltage of +40 kV, recognizing that the closing can take place at any point in the cycle b. The time it will take for C to reach this voltage c. The peak current during the operation d. The steady state rms current passing through the switch after the transient has subsided (any practical circuit would contain some damping) 3.5 —@—> Infinite A bus a Fig. 3P.4.PROBLEMS 59 A line to ground fault occurs as indicated in Fig. 3P.4, close to the secondary terminals of a 230/34.5 kV transformer. The transformer has a three-phase rating of 100 MVA; it has 0.1 pu reactance on this base. Calculate: a. The fault current b. The time to peak of the transient recovery voltage when the circuit breaker opens to interrupt the fault current. A value of 12.7 nF can be assumed for the effective capacitance per phase of the transformer secondary winding. 3.6 Fig. 3P.5. The capacitor C, in Fig. P3.5 is initially charged to Vc,(0); C, is uncharged. Show that when the switch is closed, the voltage across C, is of the form Vo,(t) = AVe,(0){c0s «44 ~ cos wf} Determine the constants A, ,, and w,. 3.7 X,, = 0.008 pu X,, = 0.00015 pu Xo, = 3.2 pu Xo, = 4.2 pu Fig. 3P.6. The circuit in Fig. 3P.6 represents one phase of a three-phase installation in which capacitors can be connected to, or removed from, a 230kV bus. The reactances are based on 250 MVA, three- phase. The switch S$, has been closed for some time. The switch S, is closed when the supply voltage of this phase is 20° beyond its peak3.8 3.9 3.10 SIMPLE SWITCHING TRANSIENTS and C, is completely discharged. Compute the lowest voltage and the highest voltage that the point P attains during the transient disturb- ance that follows the closing of S,. A 138kV/13.8kV, 20 MVA, three-phase transformer has a reactance of 10% and a resistance of 0.4%. Calculate to a reasonably close approximation the peak fault current in the low voltage winding, under the worst conditions, if a three-phase fault occurs at the low voltage terminals. For the purpose of the calculation the impedance of the source can be considered negligible. Tf the fault current is interrupted by a circuit breaker on the high voltage side of the transformer, and if the capacitance per phase of this winding is 5200 pF, determine the frequency of the transient recovery voltage seen by the breaker. Refer again to Problem 3.8. Let us now assume that the impedance of the 138 kV source is not negligible and that it has the same X/R ratio as the transformer itself. Let us further state that a symmetrical fault on the 138 kV bus would develop a fault current of 18 kA rms. Determine the frequency and relative magnitude of the transient recovery voltage of the circuit breaker when it interrupts the fault described in Problem 3.2. Assume that the stray capacitance on the source side of the breaker is 12,000 pF. L Cy, Sy N | c, Fig. 3P.7. The circuit shown in Fig. 3P.7 is designed for “synthetically” testing circuit breakers, i.e., making a meaningful test on'a breaker without a power system or large test generator. The circuit breaker under test, 5,, is initially closed. The test starts by closing S, (this will probably be some kind of triggered gap), which causes C,, which has been previously charged, to discharge through S, and the reactor L. The contacts of S, are opened soon after this current starts to flow. S, arcs until current zero, at which time, if interruption occurs, the circuit automatically applies a transient recovery voltage across S). It is required to test a breaker with a peak current of 15kA at 60 Hz, and then apply a TRV with a peak of 20 kV at 900 Hz. What3.1L 3.12 REFERENCES 61 should be the values of C,, C,, and L, and to what initial voltage should C, be charged? Make a sketch showing: a, The current through S, b. The voltage on C, before and after current zero ec. The TRV If the X/R ratio of the transformer in Problem 3.5 is 12, calculate the peak value of the fault current if the fault occurs when the voltage is 70° past its peak. ©, = 100 uF; C, = 40 uF ¢ 2 L, = 10 mH; L, = 20 mH Fig. 3P.8. C, and L, in Fig. 3P8 are oscillating; C, and P, are de-energized. The switch is closed when V,, = 0 and J, = 500 A. Calculate: a. Peak voltage reached on C, b. Peak current reached in L, c. Frequency of the current in L, d. Maximum voltage on C, prior to closing the switch REFERENCES 1. S. S. Mikhail and M. F. McGranaghan, “Evaluation of Switching Con- cerns Associated with 345kV Shunt Capacitor Applications,” Trans. IEEE, Vol. PWRD-1 (1986), No. 2, pp. 221-230. 2. T. M. McCauley, D. L. Pelfrey, W. C. Roettger and C. E, Wood, “The Impact of Shunt Capacitor Installations on Power Circuit Breaker Applica- tions,” Trans. IEEE, Vol. PAS-99 (1980), pp. 2210-2222. 3. R. P. O'Leary and R. H. Harner, “Evaluation of Methods for Controlling the Overvoltages Produced by Energization of a Shunt Capacitor Bank,” CIGRE Report No. 13-05 (1988). 4. R. H. Park and W. F. Skeats, “Circuit Breaker Recovery Voltages, Magnitudes and Rates of Rise,” Trans. AIEE, Vol. 50 (1931), pp. 204-239.4 Damping 4.1 SOME OBSERVATIONS ON THE RLC CIRCUIT All LC circuits that we have examined have been loss-free in that no dissipative element has been included. But all practical circuits have losses arising primarily from circuit resistance and iron losses in equipment. In addition, the system loads represent very important dissipative elements. Whether loads or losses, the dissipation is accommodated by including resistance in the circuits. In making transient analyses, all losses usually are neglected in the first instance, greatly reducing the complication of the calculations. Moreover, this approach leads to solutions that give severely high overvoltages. Once the general behavior of the circuit has been established, the modification introduced by the system losses can be consid- ered separately. Introducing resistance always has the effect of damping out the natural oscillations of a circuit. How quickly this occurs, or indeed whether the circuit can oscillate at all, will depend upon the extent of the losses, or, put another way, the value of R relative to the values of L and C. This chapter is devoted exclusively to studying two very important circuits, the parallel and series RLC circuits, which are illustrated in Fig. 4.1. These circuits are considered important because the networks involved in many practical transient problems in power systems can be safely reduced to one or the other of these configurations for the purpose of analysis. In many other instances, the network can be reduced to a number of these simple circuits which are so loosely coupled that, on being treated indepen- dently, they yield results of acceptable engineering accuracy [1]. The behavior of electric circuits is surprisingly simple. This is especially true of the two circuits now under consideration. The differential equations describing the two circuits in their transient state are essentially similar. For the parallel circuit, Fig. 4.1a, the equation can be written as Oo 1 dd, © Ee tRO a tie FO (4.1.1) where ® can be the current in any of the branches, or the voltage across the circuit. The F(t) depends upon the drive. For the series circuit, Fig. 4.1b, the equation is ay Rdb, wo tL at TET FO (4.1.2)SOME OBSERVATIONS ON THE RLC CIRCUIT 63 | (a) (b) Fig. 4.1. Parallel and series RLC circuits. where w is the voltage across any component or the current through the circuit. We note that the only difference between Eqs. 4.1.1 and 4.1.2 is in the coefficients of their second terms. These coefficients are themselves interesting. To satisfy Eqs. 4.1.1 and 4.1.2 they must have the dimensions of T~'. Accordingly we designate them as follows: parallel circuit time constant = RC = Tp (4.1.3) ee L series circuit time constant = & = T; Observe that the product of these time constants is the square of the angular period of the undamped circuit: T,T,=LC=T? (4.1.4) If we define a parameter 7 as the ratio of the resistance R to the surge impedance, Z, =(L/C)', that is, —— r(£) (4.1.5) Zy then the quotient of the time constants 7, and 7; is equal to 77: ip RiGee T; ~L” (4.1.6) These relationships lead to a duality in the analysis of the series and parallel circuits. While the basic phenomena in these two circuits are simple, the solutions to the equations can be complicated. There is ample scope for error in manipulating the algebra. To reduce the computational labor once the fundamental ideas have been grasped, we use the following approach to short cut several stages of the normal analysis of a problem.64 DAMPING There is a small number of transforms that appear regularly in the operational solutions for problems involving these circuits. We shall invest the effort required to find their inverse transforms and then we shall plot these inverse transforms in a series of dimensionless curves using 7, defined in Eq. 4.1.6, as a parameter. Thereafter, when we encounter these trans- forms in solving practical problems, the solutions can be extracted from these curves with about the same effort as one would expend in using a graph of a trigonometric function. 4.2 THE BASIC TRANSFORMS OF THE RLC CIRCUITS Before proceeding to the general approach just described, we look at a few specific problems. This will help verify some of the assertions that have been made and introduce the basic transforms referred to in the last section. Consider the solution for the current in the inductor in the paralle] circuit, when a switch is closed in the capacitor branch, allowing C to discharge through R and L. Let this current be /,. The current from the capacitor must be equal to the sum of the currents in the other two branches. Calling the capacitor voltage V,, (4.2.1) We may also write di, dt (4.2.2) Eliminating V, from Eqs. 4.2.1 and 4.2.2, ai, L dl, TOL Ge TR te or J a, RC dt or di, id, I, e + Fr nl + 3 =0 (4.2.3) which accords with the assertion made in Eq. 4.1.1. Here F(t) is zero because there is no external drive. In the symbols of the Laplace transform, Eq. 4.2.3 may be writtenTHE BASIC TRANSFORMS OF THE RLC CIRCUITS 65 (2 + re + A): i,(s) = (s+ Fe) u00+ 1;,(0) (4.2.4) But 7,(0)=0, and from Eq. 4.2.2, 1;(0)=V.-(0)/L. With these substitu- tions in Eq. 4.2.4, eee (s/Tp) + (1/T*) i,(s)= —— (4.2.5) Equation 4.2.5 contains the basic transform of the parallel RCL circuit. It, or variations of it, always appear in the operational solution when such a circuit is disturbed. The transform is evaluated in the following manner: ee 8 +(s/Tp) + (IT?) (8—5,)(8— 5) _ ( 11) (s,—5,) \s—s, where s, and s, are the roots of s* + (s/Tp) + (1/T”) =0 and are given by aoe eee “oT, InP --L-1(4 4)" “9 “or, Wr P Thus 7. V,(0) eee WO" pamewrT leas Teal 2 We recognize a simple transform here and can write the inverse transform readily: ¥e(0) as (67 - 8) (4.2.8) Liy= —— Ve) O> Tarr) - ary The form of this solution will depend upon the values of s, ands. From Eq. 4.2.6, if 1/7} >4/T’, then s, and s, are real. If, on the other hand, 1/T; <4/T”, then s, and s, are complex. These conditions can be expressed in terms of the parameter », defined in Eq. 4.1.5: R 1-7 (4.2.9) Thus, if 1/13 >4/T?, then 7<1/2; conversely, if 1/72<4/T?, then 71> 1/2. Consider the case where the roots are complex. Rewriting Eq. 4.2.6 in terms of 7,66 DAMPING 1 ; ! ~ yp, [= ian 1)" i” ; ; (4.2.10) __ An? — 1)" > Fp Nt in So that for the condition where 7 > 1/2, the solution (4.2.8) can be written V_(O)tpe (77? i4y = 1) | aw "| Tenet [oe ag or LO = or (4.2.11) V0): 2 tremaees LOA FZ Gat Ty si -1) v2 _t For the inverse of the basic transform with which we started (Eq. 4.2.5), we can now write 1/2. Let (1-4n7)'? =a", and let t/2T, =x 3/23 95/245 (n= H22n-1 Beets a Geetes auex a “x = + to tet sinh al x= a2 + E in oh al? 3 ays meet Et (4.2.15) as 9 1/2, a'!?-+0. Therefore as 7 1/2, sinh a'?x 2 a Thus 1 i 1 wry, gg Cette tL {s+ (1/2Tp)° ’ 2T, = te 2%P (4.2.16) There are therefore three solutions for the current in the inductor of a parallel RLC circuit when the capacitor is discharged through the other two branches. The form of the solution depends upon the value of 7. These solutions are collected together again here for convenience: V(0) 2Tpe (77 ' 1Q=-¢ Ge ary 0? -1)' eons from (4.2.11) V0) et HOE z VA) eotitte from (4.2.16) wate LW- Ka 21'n€ inh (1 — 49 py? SF from (4.2.14) a Examples of each solution are shown in Fig. 4.2. Suppose now that instead of the inductor current J, we wish to find the capacitor voltage V,. Then starting with the same Eqs. 4.2.1 and 4.2.2 we eliminate the current. Differentiating Eq. 4.2.1: dV Ldh, , 1 We dr? °C dt’ RC dt 0 (4.2.17)68 = DAMPING 2Tp Vel) Lay? = 1) A(t) { a>t Underdamped 7 >t n=t Critically damped n<} 7 Over damped +t Fig. 4.2. The inductor current in a parallel RLC ciscuit for different degrees of damping. Substitute for d//dt from Eq. 4.2.2, aV. 1 ao, Ve “aw 'RC a tIc~° or (4.2.18) av. ea Ve a a tpn This is identical in form to the equation for current. Transforming Eq. 4.2.18 gives (+ = + favo (s+ F)veO+¥e0) (4.2.19) It is evident from Eq. 4.2.1 that 1 _ Vel) Ve) =~" RETHE BASIC TRANSFORMS OF THE RLC CIRCUITS 69 1(0) _ Ve(0) Therefore Vi(0) = - 4 ~ “Re but [,(0) =0, so : Ve(0) _ V0 Ve(0) =~ Re = oo) (4.2.20) When this is substituted in Eq. 4.2.19, it cancels with part of the first term on the right and gives for the operational solution s v¢(s) = V-(0) F 36/7) GIF) (4.2.21) Equation 4.2.21 contains the second important transform associated with the parallel RLC circuit: s s? + (s/Tp) + (A/T?) This is the transform of Eq. 4.2.1 multiplied by s. It is apparent from Eq. 2.2.12 that multiplying a transform by s has the effect of differentiating inverse transform: s ef4g1_ 1 S'+(s/T,)+ (A/T?) ats? + (S/T) + (L/T*) + F(O) (4.2.22) Equation 4.2.22 allows the transform to be evaluated at once from the first basic transform. For n>1/2: ge =—__+_ s?+(s/Tp) + (A/T?) inant 1)" -rordcgs (ag? = 82 5h = EAD | ‘ 27, Gy? 27, For 9 = 1/2: gi + ___- “omy a 4.2.23 iPeantor a) on For 7 <1/2: 5 s? + (s/Tp) + A/T’) L2Tp aya _ sinh (1 — 4n°)'"? x} {cosh (1/49?) it, sayy 2,70 DAMPING Fig. 4.3. Exciting an RLC circuit with a ramp of current. There is one other important transform that is frequently encountered with the parallel RLC circuit, and that is 1 s[s? + (s/Tp) + (A/T*)] As an example of how this can arise, consider what happens if such a circuit is excited by a ramp of current, J's as shown in Fig. 4.3. The current increases uniformly at a rate of /' A/s. We assume the circuit is initially dead, that is, there are no currents flowing in any branch and C is discharged. After the ramp of current is applied, the currents in the three branches are: zI< In the resistor: In the inductor: > ih Vdt In the capacitor: cf — Their sum is the injected current [’t: ¥ +o%s4 t [vara or, differentiating, av ela # (4.2.24) pe ad? T, Taking transforms, (* + oF + Aa) = 4 + (5 + 7") +V"(0) (4.2.25)THE BASIC TRANSFORMS OF THE RLC CIRCUITS 71 Now, V(0)=0@ and V’(0)=0 also, since V’(0) = I¢(0)/C and initally there is no current in any branch of the circuit. Equation 4.2.25 therefore reduces to 1 v ~ s+ G/T») +G/T)] © (4.2.26) v(s) This is the third transform we spoke of. This transform can be evaluated with the aid of Eq. 2.2.16, which says that to multiply a transform by 1/s is to integrate its inverse transform. So the inverse transform of Eq. 4.2.26 can be obtained by integrating Eqs. 4.2.12, 4.2.14 or 4.2.16, depending on the value of n. Alternatively, we can take partial fractions: 1 s[s? + (s/Tp) + Q/T’)] _ op. A s _ 1 =r; P46) +h) aera! (4.2.27) This inverse transform of each term on the right has been obtained before. The result is as follows: For 7 > 1/2: von[ si (4-1)? (477-1)? 27, 1 po ef fe tet) +a ri ae ve eT +cos (4 — 1) Tr, For = 1/2: ge — =(2T, yt ~et7(1 4 x)I s[s+(1/2T,)} “ 2T; For 7< 1/2: = 1 +f _var,] sinh (A—4y7)'? ot erf-oen 2° FF @Tp) +a) . (1-47)? 2; - aye] + cosh (1 = 4y?)!"* 57 (4.2.28) We now have expression for the inverse transforms of the three basic transforms2 DAMPING 1 s : 1 _ 8 +(s/Tp)+(1/T?)’ 5? +(s/Tp) +(1/T*)’ s[s? + (s/Tp) + (/T?)] of the parallel RLC circuit. As predicted in Section 4.1, the expressions are complicated. To simplify we reduce the solutions to sets of dimensionless curves for different values of 7. 4.3 THE GENERALIZED DAMPING CURVES The best approach to this subject is to consider a specific example. In Eq. 4.2.11 it was shown that when the capacitor in a parallel RLC circuit is discharged through the other two branches, the inductor current is given by = ap sin (a — 1)" (n> 1/2) (4.2.11) We observe that wherever f appears, it appears in conjunction with the time constant T,. Let t'/2n = t/2Tp, so that t=eten (4.3.1) where, as before, T is the angular period 1/w, of the undamped wave, and time is therefore measured in units of this amount. Substituting Eq. 4.3.1 in Eq. 4.2.11, V(0) 7, 2ne” nt L()=— 7 Gn? oy sin (4n? — 1)" oA (4.3.2) The first part of this expression, V(0)7/L, can be written V(0)Z, inasmuch as T/L =(LC)'/L =1/Z,. As expected, it has the dimensions of current. The remainder of the expression, nett arp? sin (49? — 1) is dimensionless, being a function of the ratios 7 and t’. Consider the situation where there is no damping, which is reached in the limit as 7 ~. For this special case Eq. 4.3.2 reduces to V(0) % at 2n 1,(0) = sint’ (4.3.3) which is a sinusoidal current with a peak amplitude V(0)/Zy. For any finite value of 1, sin t' must be replaced by the functionTHE GENERALIZED DAMPING CURVES 73 2 tt fila) = 7 sin (4n? ~ 1) ao (4.3.4) A family of curves is drawn for f,(n) in Fig. 4.4 for different values of 7 with 1’ as the independent variable. These curves can be used to evaluate 1 = 8° +(s/Tp) + (1/T’) whenever this transformer occurs. All that is necessary is to compute the value of 7 for the circuit being studied. Fig. 4.4. Plot of the function NON gyre sin?! 7F 2 being lon 1 R ( i. ¢' =_____ where n=5=R 7? Gam GiTA) = (VT ize "2 For series circuits, substitute A for 7.74 DAMPING wae Fig. 4.5. Subsidence transient in a parallel RLC circuit. The peak of the undamped wave sets the vertical scale for the curves, and knowing 7 we have determined which curve represents our solution. Equa- tion 4.3.1 simplifies the obtaining of a time scale. An example will help to clarify this procedure. Figure 4.5 illustrates the problem discussed at the beginning of Section 4.2. In that section an expression was derived for the current in the inductor when the switch S is closed. We will now make use of that analysis and the dimensionless curves of Fig. 4.4 to obtain a numerical solution for the values shown in Fig. 4.5. The operational solution for 1,, was given in Eq. 4.2.5: V-(0) 1 i= (S/Tp) +(IT") s The structure of the transform tells us that our curves in Fig. 4.4 are appropriate. The first step is to calculate the surge impedance of the circuit: L\"? (8x 1073\"2 zo-(b)" =(%2)"" 209 Next, obtain 9: R _ 430 _ 7,7 33 72 This means that the curve identified as 7 = 1.5 in Fig. 4.4 closely represents the form of the current /, . Without any damping the current peak would be Ve(0) _ Zo Because of the damping introduced by R, the curve for = 1.5 peaks at only 0.65 of this value: Z,(peak) = 70.7 x 0.65 = 46.0 ATHE SERIES RLC CIRCUIT 75 To obtain a time scale, we must find the angular period T: = (LC)? = (8 x 107? x 1077)! = 28.3 ws This means that each unit of ¢’ in Fig. 4.4, for this specific problem, represents 28.3 js. This means that J, reaches its first peak after approxi- mately 38 ys, and the frequency of the damped oscillation is approximately 5306 Hz. The numerical answers to this problem were gleaned directly from the generalized curves, without recourse to evaluating or even reaching the expression for I, derived in Eq. 4.2.11. In a very real sense, the curves of Fig. 4.4 are a graphical representation of the expression i 1/ 2. ed (s/Tp) + A/T?) There is no need to evaluate the transform for each problem, since this has been done. We now go directly from the operational solution of a problem to the solution proper in graphical form. In precisely the same manner, and with exactly the same substitutions (letting 1/27; = 1'/2n) that were applied in developing the curves of Fig. 4.4, the inverse transforms s a 1 d £ PL 7 s[s’ + G/T,) + ( go ——______ 8? +(s/Tp) + (1/T?) can also be represented by sets of dimensionless curves for different values of . This has been done in Figs. 4.6 and 4.7. We shall make extensive use of all these curves in later chapters of this book. In the meantime we direct attention to the RLC circuit shown in Fig. 4.1b. 4.4 THE SERIES RLC CIRCUIT At the beginning of this chapter, it was asserted that the basic equations describing the behavior of parallel and series RLC circuits were identical in form, as given in Eqs. 4.1.1 and 4.1.2. An example will now be taken to demonstrate this point. Suppose a battery of voltage V is connected to a series RLC circuit in which the capacitor is initially discharged. What current will flow? Let the unknown current be J. Adding the voltage around the circuit leads to the following equation: dl uf) _ IR+L 9 +a] dav (4.4.1)76 = DAMPING Differentiating Eq. 4.4.1 and rearranging, a1 Rd Tt _ o a@tLatie-° (4.4.2) This is indeed of the same form as Eq. 4.1.2. Introducing the series time constant L/R = Ts, Eq. 4.4.2 can be rewritten as di id st @ hap (4.4.3) Transforming this equation, 2i(s) — (0) — (9) + 1). — 1. *i(8)— s1(0) — (0) + = 0 or s)= (s+ 4) +1'(0) (4.4.4) Ts Inasmuch as the circuit is inductive, the current must start from zero when the battery is connected, therefore (0) =0. Also, from physical reasoning, it is clear that /'(0) = V/L, since at the moment of closing the switch, there is no voltage on the capacitor, and /R = 0, therefore the only term in Eq. 4.4.1 is the second. Inserting this information into Eq. 4.4.4, v 1 L s? +(s/T,) + (/T’) (4.45) i(s)= This is the same basic transform that we obtained from the parallel RLC circuit. The form is precisely the same, the difference in detail is the appearance of T, rather than T, in the coefficient of s. The inverse transform can be derived by the same steps that were used with the parallel circuit, and perhaps not surprisingly, the results are similar. For the oscillatory condition, _~V__ 2Ts “12Ts g 2 yin _t_ =D ay sin 4a? (4.4.6) This is the same as /, in the parallel circuit with the substitutions of 7, for T, and A for 7. The parameter A is the reciprocal of 7, that is, A=— = (4.4.7)THE SERIES RLC CIRCUIT 77 Similar expressions for the current can be developed for the critically damped and overdamped conditions: I= Fev (A= 1/2) (4.4.8) I= tac P% sinh (1-442)? 5 £ (a<1/2) (4.4.9) We note that for critical damping A=1/2 or R=2Z). If R<2Z, the circuit is oscillatory, if R >22Z, the circuit is overdamped. Contrast this with the parallel circuit. Here for critical damping n= 1/2 or R= Z,/2. The circuit is oscillatory if R > Z,/2; it is overdamped if R < Z,/2. This interest- ing duality extends much further. It is clear from the example just studied that For A> 1/2: “1 1 Ces v2 _t go! ~——___, = 88 sin (4a? -1 FEIT) +Q/F) © Gat aay? inl ) For A=1/2: 7 1 “unt, inane 4.4.10) panto“ (4.4.10) For A<1/2: 7 : ee FGI E OTP are w= AY a With the substitution of A for and T, for T,, these are identical to the corresponding inverse transforms for the parallel circuit set down in eqs. 4.2.12, 4.2.14, and 4.2.16. Consequently, making the same substitutions, the generalized curves of Fig. 4.4 developed for the parallel circuit can be used equally well for the series circuit. By an extension of the same reasoning, this is also true of the curves of Figs. 4.6 and 4.7. For the series case t That is, as before, time is measured in units equal to the period of the undamped wave. We will now illustrate how the generalized damping curves are used to solve a practical power system problem.ha 7&8 DAMPING a \ _ _ -8 \ res Le <09 : nie + | = Fig. 4.6. Plot of the function -ran v2 sin(4n? - 1)'7¢"/2n fn) =e [coscan? ee Crs being s ryan) wae a For series circuits, substitate A for n.THE SERIES RLC CIRCUIT 79 : RY | : fa - / ; a |_| ett ib : Ih 7 Mh Za a ,—] Fig. 4.7. Plot of the function fan={1-e| sin(4n* — 1 2 costn? — 1°29} Gaza? being 1 g- 1 R (¢ a ae h eee a gaan, eae ae i) a80. DAMPING 4.5 RESISTANCE SWITCHING Many circuit breakers, especially those used in transmission circuits, employ resistors in the course of their operation [2]. Their primary use in modern circuit breakers is to reduce voltage transients on closing, a subject that is discussed in chapter 9. However, they were used originally during opening [3] and still provide a useful function at these times, when they are present. Such resistors serve one of two functions. In a multibreak circuit breaker they may be used to help to distribute the transient recovery voltage more uniformly across the several breaks. Alternatively, their purpose is to reduce the severity of the transient recovery voltage at the time of interruption by introducing damping into the oscillation. A resistor of comparatively high ohmic value would suffice for the first of these duties. The only requirement would be that its resistance be low compared with the reactance of the capacitance shunting the breaks at the frequency of the recovery transient. To reduce the transient recovery voltage requires a considerably lower value of resistor. Figure 4.8 shows a typical resistance switching circuit reduced to its barest essentials. In this figure, L represents the system inductance, C the stray capacitance shunting the breaker, and R the resistor used to modify the recovery transient. When the fault current proper has been switched, a residual current will remain flowing through R. This must be interrupted subsequently by opening the auxiliary interrupter S. We now show how to determine the value of R to achieve a desired modification of the transient recovery voltage in any particular situation. It was explained in Section 1.5 and illustrated in Fig. 1.4 how the principle of superposition can be used to calculate transient recovery voltages. This is done by determining the response of the circuit to a current equal in magnitude but of opposite sign to the fault current, when it is injected into the circuit at the switch contacts. Viewed from the switch contacts, the circuit elements R, L, and C appear in parallel. Let us suppose that the fault current being interrupted is symmetrical. It will be given by Viol. The problem thus reduces to that represented symbolically in Fig. 4.9. The transient period of interest is usually short compared with the time for a half cycle of the injected current wave. This is because the natural frequency of L Circuit breaker “SOTO x source @) x rau co x. Fig. 4.8. Circuit breaker, with shunt resistor, clearing a fault.RESISTANCE SWITCHING 81 Fig. 4.9, Equivalent circuit for the resistance switching problem. the circuit is usually much higher than the power frequency. For this reason, the injected current can be treated as a ramp, of slope V/L amp/s or ide (4.5.1) where V is the instantaneous system voltage at the moment of interruption. This is precisely the same circuit and the same stimulus as we examined in Section 4.2 and illustrated in Fig. 4.3. It was shown there that the oper- ational solution for the switch voltage was given by 1 v u(s) = 5s? G/T) + G/F) © (4.2.26) in which J’ = V/L. The inverse transforms for this expression were given in Eq. 4.2.28 for different values of 7, but, more important, their form was shown in the generalized curves of Fig. 4.7. The (1 — cosine) wave, for which 7=®, is the unmodified, undamped recovery transient. The other curves show the effect of introducing various values of shunting resistor, which the different values of 7 imply. A specific example will make this clear. Consider a 345 kV, 25,000 MVA, three-phase air blast circuit breaker. It is capable of interrupting approximately 40,000 A, which might be accom- plished by having two breaks or interrupters in series. Without damping of any kind, interrupting a symmetrical current, it was shown in Section 3.3 that the transient recovery voltage would reach a peak of twice the system voltage. In this example, this would be 2 x 345V2/V3 = 562 kV, for a system with grounded neutral. Suppose that it is desired to reduce the voltage to 70% of this value or approximately 1.8 for 7. Before this can be translated into a specific ohmic value for the resistor, we must know more about the circuit. A 345 kV system with an available fault current of 40,000 A has a 60 Hz reactance of 345,000/40,000V3 = 5 Q, which represents an inductance of 13.2mH. This then is the L in Fig. 4.9. A typical value for the capacitance C of a 345 kV bus to which the breaker is connected might be 25,000 pF. This gives, for the surge impedance of the bus,82. DAMPING [ee 1.32 x 1077)" _ 2-(¢ (35 eee Now, if n= 1.8, R=nZ, = 1.8X 725 = 13000 Thus, to achieve the required reduction of transient recovery voltage, each of the two interrupters in the circuit breaker should be shunted with 650 0. Figure 4.10 shows an actual resistor designed for this kind of duty. This has a lower ohmic value so that it would effect an even greater reduction in the transient voltage peak in the situation just examined. Note how the resistor is fashioned to reduce its inductance. Any inductance in the resistor will reduce its effectiveness. Accordingly, it is formed from a thin ribbon of stainless steel. The ribbon goes back and forth between the inner and outer radii, progressing around the circle. Alternate layers are progressed in opposite directions so as to reduce the inductance still further. When the resistor current is subsequently interrupted a second transient will be initiated. To study this it is necessary to introduce the capacitance C’, shunting the resistor break, and, if it is significant, the inductance of the focal loop to the fault. This is shown as an equivalent circuit in Fig. 4.11.LOAD SWITCHING 83 L R s TOC AN x. Q == C . VOTO x Fig. 4.11. Resistance switching-—equivalent circuit for interrupting the resistor current. This appears to be a more complicated circuit than we have analyzed so far, but it can be solved along similar lines. 4.6 LOAD SWITCHIN The most frequent functions performed by some switching devices are to switch on and switch off loads, which in many instances can be represented by a parallel RL circuit. Low power factor loads will be predominantly inductive, high power factor loads predominantly resistive. When such a load is switched off, the effective capacitance of the load becomes important in determining the form of the transient generated. This is illustrated in Fig. 4.12. It is obvious that this resembles the resistance switching just described (cf. Fig. 4.9). The load depicted in Fig. 4.12 has a relatively high power factor. When Source Load (a) 4.12. Transient evoked when switching a load, (a) Simple equivalent circuit. (6) Transient voltage across the load. (c) Transient voltage across the switch.84° DAMPING the current extinguishes, the instantaneous voltage, and therefore the voltage across the load, is V,. Now C will be charged to this voltage and will subsequently discharge through L and R. In Fig. 4.12b this is shown as a damped oscillatory discharge and is in fact a damped cosine wave, the form of which can be extracted from Fig. 4.6 for the appropriate value of 7. ‘The effect of power factor is interesting to observe. As the power factor improves, the current comes more and more into phase with the voltage, so that V, diminishes. At unity power factor (purely resistive load), voltage is zero when current is zero, so there is no transient at all. The situation is different if the power factor is corrected to unity. An example will make this clear and also provide an opportunity to look at series damping. Are furnaces are commonplace in industry. Ore is smeited and metals are alloyed and refined by melting them with the intense heat of an electric arc, using graphite electrodes. Such installations usually operate at a low voltage and high current and are consequently fed by a step down furnace trans- former. They are characterized by low power factor and frequent switching. Capacitors are frequently connected to the high voltage bus to improve the power factor; they are switched with the transformer and furnace. Fig. 4.13 shows one phase of such an installation, with its equivalent circuit. Delta and wye connections can be used; the figure shows one phase of a wye connected circuit. The following details apply: Transformer: 60 Hz, 13.8kV, 20 MVA Y/Y connected and solidly grounded. When fully loaded at rating, power factor = 0.6. C corrects the power factor to unity, as seen at the supply bus. Our present purpose is to investigate the transient evoked by switching off the fully loaded transformer. 20.000 Li _—— S oad current = 75-73 = 836.7 A= 1 4 13.800 5 Total Z = =~ =9. a .6 = 53.13” ‘otal Vi ge7 795222, b= cos '0.6= 53.13 Total R = 9.522 cos ¢ =5.7132=R, +R, Total X = 9.522 sin d = 7.61802 = X, + X, L=20.2mH When the current /, is interrupted at current zero, the currents /, and / are equal and opposite. We measure time for the subsequent transient from this instant, therefore 1,(0) = ~1(0) = 836.7 sin d = 669.4 A V,(0) =0 (4.6.1) 3.800V2 je = 16.3.0 X= Rx 669.4LOAD SWITCHING 85. Transformer Transformer oe Load p.f. Correcting = Capacitor Furnace @ () 4.13. Arc furnace installation, (a) schematic, (b) equivalent circuit. whence C= 157.6 pF Steady-state conditions are depicted in the phasor diagram of Fig. 4.14. The post interruption transient can be computed from the circuit shown in Fig. 4.15, with the initial conditions given in Eq. 4.6.1. If the item of interest is the current, we write @I oid, t_ wt atp? (sy(s2+ 2+ L)= it a its)(s + T, + A )=(s+ 7) 0) 20) (4.6.2) Fig. 4.14, Phasor diagram for loaded arc furnace.86 = DAMPING L = 157.6 pF Cc 0.2 mH R=5.7139 R Fig. 4.15. Equivalent circuit for the switch- ing transient. It is also evident that dl Lat iRaV and in particular LI'(0) + 1(0)R = V-(0)(=0) so that Ko) Ts 1'(0) = -1(0) z 7 Substituting in Eq. 4.6.2 gives s + (S/T) + (L/T?) KO) i(s) = which directs us to Fig. 4.6, but first we must calculate A. : Ly 20.2 yr. a=(k =(sq573) = 13 Zo _ 143 _ Rig 593 go We can identify this curve from Fig. 4.6, which indicates that the current starting at —669.4 A, swings to a positive peak slightly in excess of half of this value, a half cycle later, and then to about —0.23 x 669.4 A after a further half cycle. To compute the transformer terminal voltage we write #Ve A We , Ve = aft de iT; at? vels)(s? + zt F)=(s+ F)ven)’? + VEO) =0 but Cd. dt =-1 80 Ve(0)=DAMPING AND FREQUENCY 87 1 (0) S+(/T,)+Q/T?) C (4.63) therefore v¢(s) = For this we need Fig. 4.4, and again the curve for A = 2, which tells us that the peak voltage reaches about 72% of the undamped value, [-1(0)/C]T, or —J(0)Z,. The first voltage peak after current zero is therefore 0.72 x 669.4 x 11.43 = 5.509 V since 1(0) = —669.4 A. To set a time scale for the current in Fig. 4.6 or the voltage in Fig. 4.4, for this example, we observe that (LC)'’? = 1.784 ms. This, then, is one division on the time scales of these figures. 4.7 OTHER FORMS OF DAMPING In this chapter we have treated damping as if it could be represented by a linear resistor connected either in parallel or in series with the L and C branches of a circuit. This is an adequate representation for many practical situations, for although the energy may be dissipating in a piece of neighbor- ing steel work, as a consequence of eddy currents induced therein by the magnetic field of the circuit, it is coupled into the circuit somewhat like a resistive load. There are instances where such a treatment is insufficient. These occur especially with fast transients, where the time for flux to penetrate adjacent structures, or for that matter the circuit conductors themselves, is not insignificant on the time scale of the transients. Problems of this kind are more difficult to handle. They are given special treatment as a part of Chapter 8. A practical example of this is the penetration of traveling waves into the ground. This is further discussed in Chapter 9. Another form of damping, one also associated with transmission lines and one that cannot be handled by the techniques of this chapter, is corona. Energy is extracted from the circuit in this case by currents flowing out from the conductor in small discharges into the surrounding air. This too is discussed in Chapter 9. 4.8 DAMPING AND FREQUENCY The generalized damping curves of Figs. 4.4, 4.6, and 4.7 are independent of frequency. They tell us how much damping a given value of 7 (or A) will produce per cycle of oscillation. Indeed, another way to describe the degree of damping is to express it as a ratio of consecutive peaks. If a network with a number of meshes, comprising LR and C components, is excited by a surge containing several different frequencies, the higher frequencies will88 © DAMPING damp out more quickly because their cycles elapse more quickly. It is a very common experience to see a transient which starts with several clearly defined frequencies transform with time into a single frequency transient. This effect must not be confused with physical fact that some resistive elements in power circuits are frequency dependent. For example, the resistance of a conductor in an overhead line, or of a piece of bus, varies with frequency. This is a consequence of the skin effect which causes the current to flow in greater density near the outer skin of the conductor than it does in the center as frequency increases. This topic is discussed in Section 8.4 and again in Section 10.4, which is an additional effect to the one described above. PROBLEMS 4.1 A 13.8kV, 60Hz, single phase transformer takes a current of 2.8 A rms (assumed sinusoidal) at a power factor 0.15 when energized on no load at its rated voltage. When disconnecting the transformer under these conditions, a circuit breaker chops 2 A. Calculate the peak of the voltage transient that ensues due to the chop. The effective winding capacitance is 2.5 x 107° F. 4.2 C= 32 uF L= 120 mH R= 920 V,(0) = 40 kV Fig. P4.1, The switch S, in Fig. P4.1 is closed first, followed 5 ms later by the closing of switch $,. Determine the current in each branch at the instant S, closes. Determine the current in Z at this time if S, and S, are closed simultaneously. 43 N Fig. P4.2,