The URL can be used to link to this page

Home My WebLink AboutHigh Voltage Direct Current Power Transmission 1960PE Eee ee High Voltage Direct Current Power Transmission COLIN ADAMSON AND N. G. HINGORANI © POWER SYSTEMS LABORATORY MANCHESTER COLLEGE OF SCIENCE AND TECHNOLOGY DR. UNO LAMM GARRAWAY LIMITED -- 1! KENSINGTON CHURCH STREET, LONDON, W.8 ENGLAND | } | ; First Edition 1960' © GARRAWAY LTD., 11 KENSINGTON CHURCH STREET LONDON, W.8 PRINTED IN ENGLAND BY DIEMER & REYNOLDS LTD., BEDFORD [ —r Lo a [ Ii ISt i tay [ [ f Acknowledgments This book would not have been possible without the enthusi- asm and co-operation of others. In particular the authors wish to acknowledge their gratitude to Dr. Uno Lamm of ASEA, Sweden, and to Mr. J. H. M. Sykes, Editor of Direct Current, his assistant, Miss E. Silverton, and the technical artist responsible for the diagrams, Mr. W. J. Parker. They arc also furthermore indebted to Mr. Gunnar Engstrom and other members of the staff of ASEA for much assistance in providing both published and unpublished information; to Dr. T. E. Calverley of The English Electric Company Ltd., Stafford, for his kind co-operation and for substantial con- tributions to Chapter 10; to Mr. E. L. Davey of British Insulated Callender’s (Submarine Cables) Ltd., for contributing much of the material of Chapter 14; and to Monsieur R. Tellier for kind permission to publish a recent paper as an Appendix to Chapter 14. ji The authors are also indebted to many others who have supplied material which is acknowledged throughout the text. 38 G Foreword by Dr. Uno Lamm The history of electrical power begins withdirectcurrent. The first electrical power source, the galvanic battery, delivered d.c. and the first lighting systems were fed cither from batteries or from d.c. generators, or both in combination. Soon, however, a.c. superseded d.c. in the generation, transmission and utilisation of electricity. : Why did alternating current gain such a predominant position? The answer to that question—easy voltage transformation, simple and reliable motors, general flexibility, and so on—gives rise to another question: why, then, is interest again focused on direct current? And, in turn, the reply to that query must impose the next one: Why has d.c. not been used for large power transmission in recent times? I do not intend to answer these questions here; this book is now available for that purpose. Much has been written on the subject of h.v.d.c. transmission since the early days of electrical power; the peaks in the appearance of such literature are marked by the French Thury system in the earliest years of this century, the British Transverter in the early twen- ties, the first practical appearance of the controlled mercury are converter in the early thirtics, and by the possibility becoming apparent of a practical solution to the converting problem in the late forties. It is gratifying that this literature, which has appeared in a scattered form, has been sifted and summed up in a book, and fortified by new material. Iam sure that this book will be of great help to many electrical engineers who want to make themselves familiar with the possibilitics and characteristics of h.v.d.c. transmission. But I think the sphere of interest is much wider than that. High voltage direct current power transmission, and especially the technology of controlled static converters for this purpose, is a stimulating subject of study for any electrical power engineer; many times it will inspire him to detach himself from conventional electrical concepts, and see deeper into the fundamental principles of electrical phenomena. During a time when the work of advancement of power electrical engineering is to a large extent a matter of orderly, systematic thinking and refinement of details, h.v.d.c. transmission and conversion is an area where there is still room for the romantic touch of ancient electrical enginecring. iti Kea | tecteas stan inate ts hentia nnn tn List of Principal Symbols Cy dic. line capacitance. E r.m.s. voltage between phases of the transformer secondary. Ey, Es, Es r.m.s. voltages between phases of the transformer secondary corresponding to r.m.s. voltages between phases R-Y, Y-B, B-R respectively. €1, C2, es instantaneous voltages corresponding to E, E, and E; respectively. E, r.m.s, voltages between phases of the transformer primary. r.m.s. phase voltage of the transformer primary. Ex, Ey, Eg ¥.m.s. phase voltages of the transformer secondary corresponding to phases 1 R, ¥ and B, respectively. > Eq’ nth harmonic voltage induced in a communication circuit by electromagnetic induction. Eq)” . nth harmonic voltage induced in a communication circuit by electric induc- tion. I r.m.s. value of the transformer secondary current. o d.c, line current. _ Tam maximum permissible current regulator setting. ' E Tay r.m.s. value of fundamental frequency current in the a.c. system. [ Tay Tay active and reactive components, respectively, of fundamental frequency current in the a.c. system. [ | 1, ~~ r.m.s, value of line current in the a.c. system. i Nn T.m.s. value n harmonic current. ! . ° v Toor Tao Tp, Ly, Tp Try lys Ig thy fa, fy, ig, I55 ig Toy’ Nony Lips Lia Ly, Ly oS ~ LIST OF PRINCIPAL SYMBOLS valucs of J, and J,, respectively, corresponding to y=0. r.m.s. value of secondary currents corresponding to phases R, Y and B, respectively. instantaneous currents corresponding to Ip, Jy and J, respectively. instantaneous current due to short circuiting of the commutating voltage. instantaneous currents in valves of corresponding numbers in a bridge circuit. nth harmonic currents in a communication circuit through electromagnetic and electric induction, respectively. leakage inductance of one phase of a transformer. inductance of smoothing choke. inductances of smoothing chokes of rectifier and invertor, respectively. leakage inductances of transformer phases of rectifier and invertor, res- pectively. phase number of a converter. general symbol for power rating. rating of power-factor correcting capacitor. system short-circuit capacity. power in the d.c. line. active power on the a.c. side. reactive power on the a.c. side. d.c. line resistance. , R, slope of the constant current rectifier regulator characteristic. ty Ry, Re commutation equivalent resistance of rectifier and invertor, respectively. es . ' [: vi 4 ons TTT a —S Le 7 = Nia saesacamnan ne e widge a netic <p pe per TSO ot t —— eee Yo 8A bv Py LIST OF PRINCIPAL SYMBOLS d.c. line voltage. r.m.s. value of fundamental frequency line voltage in the a.c. system. grid bias voltage. “ harmonic voltage. r.m.s. value of 7 output d.c. voltage on no-load with a=0. commutating reactance of one phase. per-unit value of X,. leakage reactance of one phase of transformer windings. reactance of compensating reactor of one phase. system reactance of one phase. delay angle of valve firing of a rectifier. advance angle of valve firing of an invertor. delay angle of valve firing of an invertor. angle of commutation or angle of overlap. angle of commutation at zero angle of delay. angle between the voltage zero and the end of commutation in an invertor. integral of voltage drop during each commutation period. deionisation time. d.c. voltage drop due to commutation. power factor angle. lagging power factor angle of an invertor. Vii | om { B, w edo a e | { ! B' Yo 8A bV ? LIST OF PRINCIPAL SYMBOLS d.c. line voltage. r.m.s. value of fundamental frequency line voltage in the a.c. system. grid bias voltage. r.m.s. value of n™ harmonic voltage. output d.c. voltage on no-load with a=0. commutating reactance of one phase. per-unit value of X,. leakage reactance of one phase of transformer windings. reactance of compensating reactor of one phase. system reactance of one phase. delay angle of valve firing of a rectifier. advance angle of valve firing of an invertor. delay angle of valve firing of an invertor. angle of commutation or angle of overlap. angle of commutation at zero angle of delay. angle between the voltage zero and the end of commutation in an invertor. integral of voltage drop during each commutation period. deionisation time. d.c. voltage drop due to commutation. power factor angle. lagging power factor angle of an invertor. vii [ 0 Sell i~- EE) . . cma as G8 Cao oa Appendix to List of Principal Symbols The existence of I.E.C. publication No. 8479) only became known to the authors after a considerable number of blocks had been made for this book. It has unfortunately not proved possible to alter these and the original symbols have been retained. The following is a key to the recommended I.E.C. symbols which differ from those used in this book. Symbol adopted E E, Ep Tay Toyo Va Vou Ye Y 8V bV/V, Ps viii Recommended LE.C. symbol SO nn ene er een nanny be panei tre, Gl B ee C Contents INTRODUCTION CHAPTER 1 GENERAL CONSIDERATION OF A.C. AND D.C. SYSTEMS.. 1.1 Overhead lines i Economic comparison—Reliability—Earth return—Stability-—Power fac- tor—Corona and radio interference—Skin effect—Tower size. 1.2 Cables 7 .: 1.3. Interconnection of different frequency systems 1.4 Generators . 1.5 Regulation .. : o. = 1.6 Limitations of high voltage dc. transmission an 7 Transformation—Reactive power—Switching—Interconnection. 1.7 Calculation of distance beyond which d.c. becomes more economical than a.c. 1.8 Conclusion .. CHAPTER 2 TYPES OF CONVERTER CIRCUITS AND VALVE CONNECTIONS 2.1 Availability of valves 2.2 Different types of valve arrangements Operation’ of different valve arrangements (Diamerral, Double star and Bridge connection)—Comparison of valve arrangements. 2.3. Transformer connections .. es . 2.4 Interconnection of different bridge units 2.5 Valve arrangement within a bridge 2.6 System arrangements of the d.c. side + : Single conductor with ground return—Two conductor system without earth. connection—Two conductor system with midpoint earthed. CHAPTER 3 BRIDGE RECTIFIER AND INVERTOR PARAMETERS 3.1 Assumptions made to simplify calculations 3.2 Rectifier parameters : Ideal conditions: zero transformer winding reactance aie no grid ‘control —finite transformer winding reactance and no grid control—Grid control, transformer winding reactance neglected—Grid control and transformer winding reactance taken into account. : ix Page amomonaonwnmn 11 11 11 15 16 18 21 24 24 24 i i | t t i t Valve voltage in a single bridge connected invertor_—Valve voltage when two bridge units operate together displaced by 30°—Effect of voltage dents—Method of compensation for the effect of one unit on the other— ( 14 [ I ; CONTENTS iI 4 page 4 3.3. Rectifier characteristics .. . a 7 . = 30 ‘ ; 3.4 Operation of an invertor .. 2 + = = = «s = 31 i 3.5 Invertor parameters and characteristics .e oe . ss . 35 ‘ i 3.6 Commutation reactance .. os . i a as 37 i 3.7 Valve voltage in bridge connected invertors : 37 | t t ' ' eT FE Compensation in a double bridge arrangement which uses one transformer r [ with two secondaries and one primary. HES 3.8 Valve voltage in the bridge connected rectifier i a . os 45 i | 3.9 R.M.S. value of the transformer secondary current .. . i i 45 [ 3.10 Converter equations in terms of the. per-unit a.c. reactances i i 47 i } 1 CHAPTER 4 ‘ BASIC REQUIREMENTS OF GRID CONTROL .. os setae : oe 49 [. 4.1 General .. a a a a a a . 49 a 4.2 Pulse method of firing = + + oo + i : 49 { 4.3. Phase shifting . = 50 3 4.4 Arrangement for converting a sinusoidal voltage into pulses of 120° ‘duration 53 |. 4.5 Arrangement for converting short impulses into pulses of 120° duration .. 55 i 4.6 Grid bias and other auxiliary supplies .. . + 56 ; 4.7 Pulse transformers. i 58 ( 4.8 Arrangements for i impressing pulses over ‘the gr id bias of indiv id ual valves i ina : F bridge connected set... i a ae . . . . 58 oh G CHAPTER 5 ; Nee COMPOUNDING AND REGULATION .. : + = .e 62 7 ft 5.1 General... i :. . . . . a Le 62 ' : 5.2. Required regulation . 7 + . . . 62 E 5.3. Invertor compounding... . ar i . -. a “ 63 ' it 5.4. Uncompounded invertor .. + . + 7 . . 66 4 5.5 Rectifier compounding .. a : . 66 a 5.6 Transmission characteristics with the rectifier and invertor compounding ‘. 71 i 5.7 Communication link . i : Pr i . + + 72 ' 5.8 Current regulation from the invertor side . + . i 73 E: 5.9 Transformer tap changing - 75 j 5.10 Regulation of converter system in relation to the requirements of thet receiving : - system... . . . = 78 B Gencral—Regulation w vhen the power supplied is only by the c converter plant—Regulation when the receiving system capacity is very large com- pared to the converter plant—Regulation when converter capacity is com- parable to the receiving system capacity. 66 73 CONTENTS 5.11 Power transmission and regulation in case of failure of the communication channel 5.12 Stability of the receiving system c 5.13. Reversing the direction of power transmission . . CHAPTER 6 PROTECTION OF H.V.D.C. SYSTEMS .. : 6.1 Basis of protection of a.c. and d.c. systems 6.2 By-pass valve 7 Re ae : : : Gencral—By-pass valve operation in a system using one ‘bridge unit at each end—By-pass valve operation in double bridge connection, both bridges on the same side of earth—By-pass valve operation in double bridge connection, with earthed neutral—Disconnection of a bridge unit. 6.3 Classification of faults ; 6.4 Faults on the a.c. side of the invertor Distant faults—Near faults. 6.5 Invertor faults : Commutation failure—Fire through or ‘grid blocking failure—Arc quenching and failure of a valve to fire—Invertor backfire—General observations. 6.6 Faults on the d.c. transmission line 6.7 Rectifier faults Backfire or arc back— Failure of a valve to fire—Fire through. 6.8 Faults on a.c. side of the rectifier 6.9 Failure of supply to auxiliary equipment 6.10 D.C. Circuit-breakers 7 : The need for circuit- -breakers—possible types ‘of dic c. circuit- breakers. 6.11 Possible solution to the problems of switching without the use of high- capacity d.c. circuit-breakers .. 6.12 Overvoltages : Internal overvoltages—External overvoltages. 6.13 Parasitic oscillations : a 7 General—Voltage steesses—Current surges—Damping methods. CHAPTER 7 REACTIVE POWER REQUIREMENTS 7.1 General Me 7.2 Uncompounded invertor .. 7.3, Compounded invertor 7.4 Sources of reactive power xi page 83 83 84 85 85 86 91 91 93 96 97 101 101 101 104 106 108 115 115 115 117 119 CONTENTS i CHAPTER 8 "page i ARTIFICIAL COMMUTATION... 14 oe sla + ee le aoe | 8.1 General considerations .. a i . 24 he a - 121 : } 8.2 Forced commutation in two steps . an An Le a +» 121 i i 8.3. Forced commutation in one step ie ale Am Pe Pe -» 124 oy 8.4 Resonant commutation .. , tt Ht tl 26 } 8.5 The use of series capacitors in transformer secondary windings se -» 127 + 3 | CHAPTER 9 T i USE OF EARTH AND/OR SEA RETURN i Aa Ae i be IT] |) i393 ' | 9.1 General .. a q i; ls a i a -- 43133 = i 7 9.2 Electrodes and their arrangement - le Be it is IT) 133 ; [ j 9.3. Nature of current distribution in the earth... aa at L. -» 136 on i 9.4 Effect of earth current on railway track signalling... i _ de) frou oa : 9.5 Corrosion caused by earth current a. Ch -. 138 [ } 9.6 Effect on the horizontal component of the earth’ S magnetic field i +.) | (439 Magnetic field change and compass error—Methods of reducing the compass error. CHAPTER 10 HARMONICS i o : “4 - i le ae Se -. 143 10.1 Harmonics on the a.c. side of a converter ee a ++ Pe - = 143 10.1.1 Current harmonics os Le tt tt -- 143 em oer Current harmonics neglecting commutating reactance and assuming perfectly smooth d.c.—Current harmonics, taking commutation reactance into account but assuming smooth d.c—Power factor—Current har- monics neglecting commutation reactance and with zero inductance. 10.1.2 Current and voltage harmonics in the a.c. system. - is .. 156 10.1.3. Disturbances created in neighbouring communication systems .. 159 Electromagnetic induction—Electric induction—Longitudinal and trans- verse voltage—Unbalance of telephone circuits—Unbalance of power circuits—Screening effects—Noise produced by induced voltages. 10.1.4 Methods of reducing harmonics on the a.c. side... : - 167 Increase in number of phases—-Effect of tuned or filter circuits—Factors : affecting synchronous machine design. 10.2 Harmonics on the d.c. side of a converter oe -» I71 10.2.1 Harmonics in the output voltage, before the smoothing reactor set | | LTE 10.2.2 Circulation of harmonics on the d.c. side, with different method of Mada i La} . | [* i connecting the bridges Pe 7 7 1 a4 ale} | | 7S i 10.2.3. Harmonics in the d.c. line ele .- 176 | in 10.2.4 Disturbances caused by d.c. harmonics and methods of reducing | | them... oe i Ps 1 v3 + ele | (LOL qj ? ' xii B | i a TS Pi ci oF — on a - a ~ a = ih [ i CONTENTS age i CHAPTER 11 page [2 | INSULATORS An ot i se a i. oP a «184 121 i 11.1 General... a : te 2 - a es -. 184 121 | 11.2 Insulation level of a d.c. transmission line a 7 a a -. 184 124 i "11.3. Insulator impedance q = as 2 ar 2 -. 186 = 126 | 11.4 Insulator contamination and wetting .. ae Zs : 2 -- = 188 127 ! 11.5 | Development of insulator breakdown .. : -. 189 ' : 11.6 Causes of surface contamination and the effect of electric field « on contamina- - tion AR : i oe = -= 190 | 11.7. Flashover and withstand voltages ‘of insulators = Pr Pe -. 195 133 | 11.8 Radio interference from insulators ile 7 oe or = -. 198 133 11.9 Stabilised and other special types of insulators - . i .. 199 133 i 136 ! [137 CHAPTER 12 138 CORONA .. 7 ar oe s i -» 200 139 12.1 Critical corona voltage and voltage stress z. +» 200 oe 12.2 General behaviour of corona discharge for positive and negative polarities 203 [ ; 12.3. Experimental results of d.c. and a.c. corona compared An a -» 206 12.4 Importance of corona losses in a.c. and d.c. transmission .. a ae 10. 12.5 Radio interference es 1 a 7 1 a os aceme4 0 CHAPTER 13 {° : HIGH VOLTAGE HEAVY CURRENT MERCURY ARC VALVES... Pr oor : 13.1 Problems of high voltage, heavy current valves _ ne re ~~ 252. 13.2. The mercury vapour are .. 7 : i ne ae -- 216 3 13.2.1 Operational features of a1 mercury arc discharge ts a .- 216 [ : 13.2.2 Voltage drop across the arc... : 220 “156 The anode voltage drop—The voltage drop in the plasma—The cathode 159 voltage drop. [: ' 13.2.3. Characteristics and mechanism of the cathode are .. : -. 221 - Current density and size of the cathode spot—Spot temperature—Vapour t jets and the tanberg effect—Erratic motion of the cathode spot—Retro- | 67 ' grade motion. = 13.2.4 Theories of spot emission 3 223 Thermionic emission—Field emission—Photo- electric emission—Theory. ‘71 ; _ of excited atoms—Combination of theories. [on 13:3 Deionisation: pen ee eee ee oS 13.4 The Grid .. p i we ds poe 175 - 13.5 Principle of high-voltage valves with voltage division a ats oo 227 [-76 13.6 Arrangement of electrodes = bs ste -. 228 13.7. Voltage dividers for valves with intermediate electrodes i ae ped eee 181 13.8 Transfer of mercury from the pool; Control of temperature and pressure... 232 Z B xiii -— t Bia i ae } i i i i 4 i i 13.9 amaaute ecient’ CONTENTS Cooling arrangements Forced air cooling—Liquid cooling. 13.10 Ignition and excitation 13.11 Gascous impurities and pressure control 13.12 Pumping arrangements .. ae 13.13 13.14 Jon bombardment and electrode sputtering Ion deficiency and spot quenching 13.15 Valve rating, arcing-back and arcing-through phenomena 13.16 13.17 Tank potential with respect to cathode Anchoring and film emission CHAPTER 14 DIRECT CURRENT CABLES 14.1 14.2 14,3 14.4 14.5 14.6 14.7 14.8 14.9 LIST OF REFERENCES INDEX Introduction : Ae Basic physical phenomenon arising in dic Cc. insulation Practical dielectrics : Impregnated paper—Polythene. Diclectric stress consideration Insulation resistance characteristics of ‘diclectrics—Stress distribution and inversion of stress with temperature—Effect of gas pressure on dielectrics. Thermal considerations and losses Cable design values y Temperature limitations —Stress values, Economics of d.c. cables compared with a.c. cables Accessories for d.c. cables General Appendix a th Introduction—Distribution of ‘Voltage i in the dielectric i in the d.c. . cable— Impregnated paper cables—Cables with thermoplastic insulation. xiv page 234 237 238 241 241 242 243 247 247 254 254 254 256 259 265 265 266 266 267 269 274 282 as acm ene rt ae art ttn ne inet i eee nee cnet peor IIS GH Har jp to Ge 2 Cem oN i” a ( | | | | Introduction References: (1) to (14) inclusive, (27) (105) and (153) The use of d.c. for the transmission of electrical energy dates from the days of Thury,“? who designed the d.c. transmission system from Moutiers to Lyons in the early days of the present century. This system had a capacity of 20 MW and extended over 138 miles, of which 23 miles was composed of cables, at a voltage of 125 kV; it operated on the basis of constant. current with several series-wound, d.c. generators in serics. Later, other systems of this type were installed. This was the period of ascendancy in the use of d.c., but the question of d.c. versus a.c. for generation, distribution, transmission and utilisation excited the wildest controversy. With increased demands for power the difficulties of d.c. generation, because of commuta- tion limitations, were realised and the lack of any means for readily changing voltage levels was a serious disadvantage. A.C. clearly appealed on account of the availability of transformers, the development and improvement of induction motors and the relatively easy way in which converting equipment could be used to produce d.c. when required. Thus progressively a.c. replaced d.c. as the demand for power increased, but the savings in transmission costs resulting from the use of d.c. were not forgotten by some pioneers, prominent among whom was J. E. Calverley. The idea was widely canvassed of a.c. generation, conversion to d.c., transmission by d.c., and conversion back to a.c. for low- voltage transmission and distribution purposes. In the early twenties a number of methods of conversion were developed, notable among which was the transverter of Calverley and Highficld in which the brushes were the only rotating parts. Its ratings were, however, inadcquate for the powers then contemplated and it was historically late in that the me.cury- arc rectifier had made its appearance. Recent work on h.y.d.c. transmission has been made possible by the development of grid-controlled, multi-electrode, mercury-arc valves which are capable of handling large powers at high voltages. The method of improving the capabilities of mercury-are valves by the use of grading-electrode structures was first patented by Allmanna Svenska Elek- triska Akticbolaget in 1929,8) and has since been further developed in Sweden and in a number of other countries. D.C. trunk-line transmission involves the erection of two converting stations with trans- formers and valves, the one for inversion and the other for rectification. From economic considerations, the high initial cost of these converting stations has to be weighed against savings in the costs of transmission; in general terms, the longer the system, the more favourable will be the case for h.v.d.c. The phenomenal increase in the extent of power systems since-the late war thus does much to explain the present widespread interest in h.v.d.c. An important additional factor is the desire to generate from water power at its source and transmit over long distances to the centres of load; the possibilities of h.v.d.c. for this purpose are particularly attractive at the present time in Sweden and the U.S.S.R. Aside from questions of long-distance transmission, there are two powerful further incen- XV + meme nine tli les Asian nS oa abort INTRODUCTION tives to the use of dc, These are the interconnection of Power systems of different frequencies, of which a good example is Japan, and the establishment of power links by long submarine cables, as in the cases of the Swedish mainland-Gotland scheme and the English Channel crossing. Also, in a country such as the United Kingdom, with dense industrial cities and relatively small available area, the purely practical consideration of in- sufficient wayleaves for more towers in urban areas demands consideration of dic. as a means for supplying the rapidly-increasing load of the future by using new or existing a.c. cables turned over to d.c. use. 7 need for water and the disposal of radioactive waste materials and effluent may demand in the main, the selection of coastal sites; it is also desirable, at any rate in the case of the Stations built in the first two decades, that they should be fairly remote from large cities. Distance is thus an important factor and so also is the desirability of running the large fry ee > we 3 2 3 a = p < oO o oO oO 5 3 co & o 5 = £ a 5 & a 2. 5 Q co Pp o oO a > Pp 5 o a Q ¢ & ° 8 oQ Ss > ° s, 3 cS g 8 5 trunk interconnection for which d.c. merits consideration. Important experiments were conducted in Germany and Switzerland both before and during the late war. In the German case an experimental transmission system of 15 MW at 100 kV, between Moabit and Charlottenburg, was built; this was intended to be a pro- totype for a 60 MW, 400 kv system of about 110 km length, part of which was built. Immediately after the war, however, the available German equipment was dismantled. The Soviet Union’s activities subsequent to this period have been extensive; an experi- mental transmission between Kashira and Moscow® at levels of 30 MW and 200 kV was put into operation in 1955, and construction work on a 600 km long, 500 MW transmission, from Stalingrad to the Donbass(!99) has been started. Sweden contributed greatly in 1954 by switching on the world’s first commercial h.v.d.c. transmission system, the Gotland system. This successful scheme®.» transmits 20 MW at 100 kV over a distance of about 100 km by submarine cable. This installation followed extensive and long-term experiments by both the plant manufacturers and the Swedish State Power Board. In the United Kingdom, after the war, investigations were mainly confined to the Electrical Research Association, Increased interest in h.v.d.c. was reflected in the decision to establish a h.v.d.c. laboratory at the Manchester College of Science and Tech- nology,2” and Preceded the important decision of the British and French Supply authorities to establish a 160 MW,-200 kV d.c. link across the English Channel, (3) (153) The subject may be said to have thirty years of recent history if one dates from the : inception of grading electrodes, and yet its basic principles do not appear in any text- ; book. High-voltage d.c. transmission is now an established fact and in the Opinion of | the authors will increase and not diminish in importance. This firmly-held opinion is the reason for this book. CoLin ADAMSON, N. G. HincoraANr, Power Systems Laboratory, Manchester College of Science and Technology. [ [ [ [ L L Ll ji L 1960. XVi | ; j i , i | { 2 a bed CHAPTER 1 General Consideration of A.C. and D.C. Systems References : (2) (3) (4) (7) (13) to (26) inclusive, (38) (43) (105) and (216). Electrical power may be transmitted by means of underground cables or overhead lines, and it is useful to compare the relative merits of a.c. and:d.c. transmission for these two cases. Considerations of this type do not settle the choice of a.c. or d.c. for any particular applica- tion, but they do indicate the economic suitability of one or the other. 1.1 Overhead lines 1.1.1 Economic comparison (a) Consider a new d.c. transmission system to compare with a three-phase a.c. system transmitting the same power and having the same percentage losses and the same size of conductor. The d.c. system is considered to have two conduciors at plus and minus of to earth, as shown in Figure 1.1. Power in the a.c. system=3 Ep. [,=/3 E, . /,, assuming that cos 6=1 Power in the d.c. system= i Vy a.c. losses=3 J,? R d.c. losses=2 1,2 R Equating line losses, 3 /,2R=2 ,?R 3 = fen Equating powers, 3 Ep. ,=1Vq : 2 Vy=3 ‘iE Ep Now assuming that the direct voltage for breakdown of an insulator is equal to the peak value of the alternating voltage to cause breakdown, insulating level of a.c. line=1/2 Ep. ky and insulation level of d.c. line a kz where k, and ky, are multiplying factors allowing for internal and atmospheric over-voltages; asa first approximation they may be assumed to be identical for both systems, i.e., ky=k2. . ~ d.c. insulation level V, 3/2 a.c. insulation level 2./2E, 2/2. V3 . = ve =0.87, Zz GENERAL CONSIDERATION OF A.C. AND D.C. SYSTEMS Thus the d.c. line will not only have two conductors instead of three (of the same size) for the a.c. line, but in addition the insulation level will only be 87 per cent of that of the a.c. line. (6) Another method of comparison is to consider an existing three-phase double circuit a.c. line in relation to its conversion to a d.c. line. Figure 1.1 Economical comparison of a.c. and d.c. transmission (i) On the basis of the same current and insulation level: Power transmitted by a.c.=2.3.E,. I, The a.c. line can be converted to three d.c. circuits, each circuit having two : V, 7 ' conductors at plus and minus 3 to earth, respectively. } Power transmitted by d.c.=3V,J, T=, | | . V, For the same insulation level oa V2E, - Power by dic. 3. Vy.ly 3.2. V2. Eye ly Power byac. 6.6.1, 6.2.1, v2 ioe J [ { [ [ OVERHEAD LINES 3 Percentage losses by d.c. losses by d.c. power by a.c. Percentage losses by a.c. power by d.c.” losses by a.c. 1 yee Thus the power transmitted can be increased by 41 per cent with the percentage line losses reduced by 29 per cent. (ii) On the basis of the same percentage line losses and the same insulation level: : V. For the same insulation level im V2E,. losses by d.c. losses by a.c. For same percentage losses —————__ = ——_____—— power by d.c. power by a.c. 612R 612R 612R 6EI, 3V lg 3.2. V2E,. Ty 1 = VU aa power by dc. 3.2 _V2.E,. V2, * power by a.c. 6E, . Ir ai Thus the power can be raised by 100 per cent on the same double circuit a.c. line with the same percentage losses. 2. (c) The calculations in (a) and (6) above have been based on the assumption that the direct voltage to cause breakdown of an insulator is equal to the peak, or crest value of the alternating voltage to cause breakdown. Recent tests seem to indicate (see Chapter 12) that this assumption is valid only for good weather conditions: in conditions of fog and atmospheric pollution, the direct voltage for breakdown may well be below the a.c. peak value. On the other hand, the actual insulation level depends also on the allowance to be made for the internal over-voltages which may be encountered during switching operation on the system, and which the insulators must safely withstand. This feature was neglected in (a) and (6) above and the factors k, and kz were taken as equal. In extra high voltage a.c. systems, the insulation level is generally based on overvoltages of, at Icast, 2.5 times the operating r.m.s. voltage; in h.v.d.c. systems, however, experiments have shown that the internal over-voltages can be reduced to less than twice the operating direct voltage. For example, the insulation -level of the Stalingrad-Donbass h.v.d.c. transmission line 95) has been fixed on the basis of over-voltages of only 1.7 times the operating direct voltage of +400 kV. Taking these points into consideration, the calculations of (a) above will now be repeated assuming that k,/k,=2.5/2 and that the direct voltage rating of the insulation will be the same at the r.m.s. value of the alternating voltage. Ratio, dic, insulation level Vu/2 . ke : a.c. insulation level — E, . ky RE Re A Tee tert ene eee ene epee mreunen ae wg 3}: - ae 4 GENERAL CONSIDERATION OF A.C. AND D.C. SYSTEMS For equal power and losses, v=3,fF - E, (from [a]) _ dic. insulation level 34/2 2 “ac. insulation level. 24/3 "2.5 =0.98 approx.=1 Thus even with the assumptions weighted against d.c., the insulation level of the d.c. line is the same as that of the a.c. line, but has only two conductors instead of the three conductors necessary for the a.c. transmission. 1.1.2 Reliability In case (b) above, since there are three d.c. circuits compared with two a.c. circuits reliability tends to be increased. If the earth is to be used in a temporary return circuit there are six d.c. circuits, and their reliability of service is increased still more. The temporary loss of one conductor results in only 17 per cent loss of transmitting capacity. In the case of the single circuit d.c. line with two conductors and earthed neutral the loss of one conductor results, at the most, in a 50 per cent loss of transmitting capacity compared to a complete shutdown of transmission in the case of a fault on any conductor of a single circuit three-phase a.c. line; thus d.c. is more reliable on this basis. 1.1.3 Earth return The fact that the earth can be used as a return conductor, whether permanently or temporarily is a great advantage. No use can ever be made of earth with a.c. systems because of the associated inductive effects, excepting when it is absolutely necessary to use the rails in electrified railways, and even in that case many precautions have to be taken. The use of the earth return in d.c. systems has limitations as well, due to possible inter- ference with railway signalling circuits, corrosion of cable sheaths, etc. (Chapter 9); but in cases where these difficulties do not arise or can be overcome, only one conductor is re- quired to transmit power. if 1.1.4 Stability It is well known that stability of an a.c. line is dependent on the power per circuit and the length of the line. For long lines it thus becomes necessary to introduce intermediate stabilising equipment such as series capacitors, shunt reactors, or even intermediate sWitch- ing stations, any of which increase the cost of the system considerably. In the case of d.c. the line length has no relation to the stability of the system; and hence it is possible to extend the d.c. line indefinitely without any intermediate compensating stations or apparatus. 1.1.5 Power factor In the previous comparison a.c. power has been taken as 1/3. E,. [, but in fact it is power to a.c. power is increased still further. V3.£,.1,.cos . Thus the ratio of d.c, To ne si pidia abies! ; -- or con is CABLES 5 1.1.6 Corona and radio interference It has been established (Chapter 11) that the ratio between d.c. and a.c. critical corona voltage must be equal to the crest factor at least, since the losses in a.c. increase much more rapidly than with d.c. at high voltages. In the comparison considered above the corona losses and radio interference may be expected to be only two-thirds of those of three-phase a.c. system, since there are only two conductors. Thus if these losses are also considered the d.c. power can be raised still further. With higher and higher voltages it has been observed that corona losses and radio inter- ference in a.c. lines rise steeply compared to d.c. and special precautions (such as using bundle conductors, etc.) may have to be adopted in a particular a.c. system whereas they may well be unnecessary in the equivalent d.c. system. 1.1.7 Skin effect Skin effect in d.c. systems is completely absent, and hence there is more uniform current distribution in the conductor, and a better utilisation of the metal is obtained. 1.1.8 Tower size The d.c. insulation level, for the same power transmission, is likely to be lower than in the corresponding a.c. case; thus the dimensions determined from purely electrical considera- tions such as the separation between conductors and the clearance to tower steelwork will be correspondingly smaller. Also, since the d.c. line will have only two conductors where three are uscd in the a.c. case, mechanical considerations will effect a further reduction. Figure 1.2 shows the results of Swedish studics™® on the tower size required for +400 kV d.c. instead of 650 kV-a.c., for the same power of 2,000 MW in each case. 1.2 Cables In the case of underground cable routes, the advantages of d.c. over a.c. are unquestion- able. Due to the unidirectional field there is practically no charging current. The steady state charging current for a three-phase cable is of the order of: 2,000 kVA/circuit/mile at 132 kV 5,000 kVA/circuit/mile at 220 kV 15,000 kVA/circuit/mile at 400 kV There is thus a distance limitation beyond which it is necessary to supply the charging current at intermediate points. This limit is as follows: 40 miles for 132 kV 25 miles for 220 kV 15 miles for 400 kV The figures in Table 1 give the losses and voltage drop of 25 c/s and 163 c/s transmission systems considered as alternatives to the transmission of 10 MW by cable from Sweden to Gotland (a distance of 100 km) as compared with the d.c. transmission system actually used.) Boa Cae | Phase distance Height of cross-arm B t 400 kVde. Figure 1.2. The relati TABLE 1 Comparison of possible methods of transmission of 10 MW from Siveden mainland to Gotland Nominal Transmission losses, percentage of Voltage drop in Type of operating power supplied from mainland cable as percentage circuit kVon — of voltage at Gotland In cable Total mainland a.c. 45 kV 17 37 17 25 c/s 70kV 10 20 7 100 kV 8 28 3 a.c, 45 kV 17 37 17 163 c/s 70kV 9 29 7 100 kV 7 27 2 dic, 100 kV 2 10 2 d.c. needs only one cable (75 mm? copper) while a.c. needs two cables of 75 mm? copper. on [: Gi cr rm! INTERCONNECTION OF SYSTEMS OF DIFFERENT FREQUENCY 7 Transmission at 50 c/s is virtually impossible. The figures in the table are self-explana- tory. They show how convenient and economical d.c. transmission is with a submarine cable, where no intermediate charging stations can be provided. By using d.c., submarine projects can be considered which would be entirely impracticable with a.c. Even on land (when transmission by cable becomes necessary due to nature of the area), the provision of the necessary charging stations makes a.c. cable systems very expensive. Apart from charging current problems, d.c. transmission by cable means the absence of ionic motion in the insulation, the absence of induced currents in the cable sheath, and the absence of skin effect. For these reasons the working stress of d.c. cables can be as high as 30-40 kV/mm, compared to 10 kV/mm in the case of stabilised e.h.v. a.c. cables. This reduces to a considerable extent the cost of d.c. cables. ; Figure 1.3 gives the approximate weights of three-phase cables and d.c. cables at different voltages for a current capacity of 5C0 A.@®) In broad terms the effective power transmitted per cable by d.c. is approximately 2} times that transmitted by a.c. 1.3 Interconnection of different-frequency systems Since the d.c. link is not a synchronous one, the regulation of the power flow can be affected independently of the frequency and voltage of the connected a.c. network. Thus systems of different frequency and frequency control can be connected by d.c. without affecting each other. @ G9 This quality will be most useful in international exchanges of power, as decided for the cross-Channel link between England and France,“) and the proposed power supply from Hungary to Italy.2® Various d.c. projects can be expected in i T i rf 100 f/f @c.3 phase &0 Ut 3e = : i eer eal _| g qe j oe TEE TTT | 0 /00 300 500 700 800kV Figure 1.3. Weights per mile of a.c. and d.c. cables - (1000 kg == 0.98 ton.) rn NE 8 GENERAL CONSIDERATION OF A.C. AND D.C. SYSTEMS future irrespective of the distances concerned, because of the required independence of frequency control. 1.4 Generators In certain cases the generators designed for hydro-electric power may, in the future, be employed mainly for d.c. transmission to inhabited areas. In these cases the design of the generator docs not have to be determined by considerations of system stability, and hence synchronous and transient reactances, short-circuit ratios and generator inertia figures may have normal values, and design may also allow for a higher power factor than normal.2) In designing the damper windings allowance will have to be made for the heating caused by the circulation of harmonic currents inherent to the working of rectifiers. On the other hand, generation is possible at a frequency different from standard frequency, which may be advantageous for hydro-electric plants with varying heads of water. 1S Regulation Due to grid control of the rectifiers, the d.c. line can be operated with constant current regulation or constant voltage regulation, both having their advantages. In fact by suitable grid control design the advantages of both systems can be obtained (Chapter 5). 1.6 Limitations of high voltage d.c. transmission 1.6.1 Transformation In d.c. there is no easy way of transformation of voltage, as in a.c. For this reason a.c. has been universally accepted for distribution. For transmission over long distances transformation could be obtained on the a.c. side of the converting stations. Thus the absence of transformation facilities is not a major disadvantage. 1.6.2 Reactive power For the operation of an invertor, it is necessary to run it ona leading power factor thereby requiring the supply of reactive power. This power has to be supplied from the a.c. side either by static or synchronous capacitors (Chapter 7). The steady-state reactive power requirements of the invertors may be of the order of 40 to 50 per cent of the real power, but with respect to transients it is advisable to have a somewhat larger figure available if required: for example about 75 per cent of the real power. Investigations into artificial commutation methods, by which invertors can be operated on lagging power factor have been done on the E.R.A. model by Busemann,®2) even to the extent of supplying the reactive power requirements of the a.c. systems. This again is considerably more expensive than a synchronous generator, and seems impracticable. 1.6.3 Switching The absence of switching facilitics is the greatest limitation of d.c. systems. With a.c., the current automatically comes to zero every half cycle, and advantage is taken of this in switchgear design: the only problem arising is to prevent restriking. nnn nti censsinee eet cnet taaninnidisntbn ye inns nein meron TE tn tt a | ECONOMICAL DISTANCE FOR D.C. TRANSMISSION 9 In d.c. practice no such current zero exists, and all the energy of the circuit has to be dissipated before interruption can be obtained by means of circuit breakers (Chapter 6.) But complcte control, including switching, can be obtained by using the grids of the converters. Switching is, therefore, not a serious problem as long as the question of interconnection does not arise. In the U.S.S.R., there.is a considerable research effort on the design of suitable d.c. switchgear?! 1.6.4 Interconnection The creation of a tee-junction on a d.c. line and the establishment of an interconnected d.c. network is severely limited until the problem of d.c. circuit breakers is solved to enable a faulty section to be isolated. This limitation can hardly be ignored since even though trunk d.c. lincs for major transmission power may be conceived initially, it is unlikely that this situation will endure with the passage of time; a.c. practice has shown that over- whelming cases become established for tapping-off and feeding-in power en route. At present reliance must be placed on the converter for carrying out switching operations (Chapter 6). 1.7. Calculation of the distance beyond which d.c. becomes more economical than a.c. The d.c. system has lower line costs than the corresponding a.c. system but needs two converter terminal stations. These cost two to three times more than the corresponding a.c. transformer stations, and thus basic economic considerations call for a certain minimum transmission distance before d.c. can be competitive. A comparison, taking into account the latest available data, has been worked out by the CIGRE Study Committee appointed for this work.23) There are three types of possible transmission arrangements: (a) A transmission system from a power station to a distribution region. (b) A connection between two existing a.c. networks. : (c) An increase in capacity of a transmission system already existing. Transformers at both ends are necessary, with a.c., in the first case only. In d.c. systems, transformers at both ends are necessary in all cases. The first case favours d.c., and has been considered for a transmission from a 20 kV generator to a 130 kV receiving system. Calculations have been made for distances from 300 to 1,000 km in the case of overhead lines and 30 to 100 km in the case of submarine cables. For single circuit overhead lines, 400 kV has been taken as a starting point, and 750 MW is considered to be the most economical load for 400 kV lines giving the lowest transmission cost per kW that can be realised with the present a.c. techniques. Against this, a +300 kV to —300 kV, 750 MW d.c. line carried on two towers with mid-point earthed has been taken as a reasonably economical choice. Considering that two-wire d.c. lines with mid-point earthed have the advantages of double circuit a.c. lines, a comparison has also been made between a 2x 750 MW, 400 kV, double circuit a.c. line and a 1,500 MW, 2400 kV dic. line. 10 GENERAL CONSIDERATION OF A.C. AND D.C. SYSTEMS In the case of cables one comparison of a power of 2x 375 MW at 2 x 300 dic. and 220 kV a.c. has been made. A voltage of 220 kV has been considered as most economical for a.c. cables. 2% 3 cables are to be used in a.c. as against 2 for d.c. practice. Three other cases of smaller capacity for submarine cables have also been investigated. Table 2 shows the cases compared and the critical distances in each case. In the case of cables, comparison has also been made when two similar a.c. systems are connected and transformers are eliminated. TABLE 2 7 Comparison of a.c. and d.c. transmission dc. Critical distance No. of valve Transmission groups No. of | Transformers Transformers type MW | kv in MW | kV | cables included excluded ; series per in ac, in ac. } per Phase ; Station | Overhead 2750 | 2x400 | 2x4 |2x750| 400 2 320 miles fe pa 2x375 | 2x300] 2x2 750 | 400 1 390 miles ify Submarine cables 2x 375 | 2x 360 2x2 750 220 2 22 miles 31 miles 7 Ps 1x 200 200 2 200 | 130 1 26 miles 31 miles o ils 2x2 200 | 220 I 19 miles 26 miles [ I at 1x100 | 200 2 100} 130 1 18 miles 21 miles Various other calculations have been made to set out the minimum distance for which d.c. is favourable. The Swiss Brown Boveri pre-war calculations? {x the distance from 209 to 400 km, depending upon power: 400 km is the figure suggested for a power of 100 MW. 7 In some circumstances the question of economic distance is irrelevant, and there are signs that legal and other non-engineering factors may become prominent in the future. For some crowded cities in Europe and North America, where wayleaves for overhead lines are becoming increasingly difficult to obtain, the case for bringing electrical power into the cities using either existing a.c, cables converted to d.c., or by new d.c. cable systems, is being actively studied. There is also a very good case for power transfer over national frontiers to be carried out by means of d.c., thereby ensuring a complete independence of control of individually and nationally-owned a.c. systems, 1.8 Conclusion et From the above discussion it is clear that d.c. transmission, though having many advan- tages, has its main application where distances are large and where power has to be trans- mitted in bulk from one place to another without interconnection and where a water barrier has to be crossed. Thus high voltage d.c. and a.c. transmission each have their own scope and fields of development, and there is no conipetition between a.c. and d.c. In fact both are complementary, and will have to be developed to take their own places in order to obtain the best advantages from available natural resources, ‘ cr? re | | CHAPTER 2 Types of Converter Circuits and Valve Connections References: (7) (10) (26) (27) (28) (29) (30) (38) and (165). 2.1 Availability of valves : The very high current density of the cathode spot in mercury arc valves, together with the high breakdown level of mercury vapour, has established the superiority of mercury arc valves for present-day requirements of power and voltage (Chapter 12). It is the development of mercury arc valves with grading electrode structures which has made possible the present developments in h.v.d.c. transmission. Semi-conductors, now being developed for high powers, do not replace mercury arc valves since they do not have the same facilities for grid control, which, besides providing the most suitable means of regulation (Chapter 6) is essential for invertor operation (Chapter 3). Mercury arc valves, involving anchoring of the cathode spot (Chapter 12), have been developed for small powers, but are not likely to be used for h.v.d.c. applications in the near future. 2.2. Different types of valve arrangements Various types of valve arrangements with different transformer windings and with half wave and full wave rectification have been adopted. These can be divided into three groups.” (a) Half wave rectification (diametral connection, which is a six-phase half-wave recti- fication, is a typical example of this). (b) Interphase transformer connection. (c) Full wave or bridge connection: 2.2.1 Operation of the different valve arrangements The manner of operation of six-phase diametral connections and double star connections with an interphase transformer is discussed in most textbooks on rectifiers; only bridge connections will be discussed here in detail. (2) DIAMETRAL CONNECTION (FIGURE 2.1) Considering the point when phase R is conducting through valve 1; after point C the anode of valve.2 (which is connected to phase B’) becomes positive with respect to the anode of valve 1. Thus commutation takes place from 1 to 2 and phase B’ conducts in the valve 2. Similarly, after point D, valve 3 takes over, and phase Y conducts in valve 3. Thus the output voltage.as shown by the thick line is obtained. The full value of direct current in the d.c. system is thus being carried by each valve over an angle of 60 degrees. Brod Tim 12 TYPES OF CONVERTER CIRCUITS AND VALVE CONNECTIONS Figure 2.1 6-phase diametral connection (b) DOUBLE STAR CONNECTION (FIGURE 2.2) In this type of connection two three-phase half wave rectifiers are working in parallel. Considering point 4, when phases R and B’ are working in parallel through valves 1 and 2 respectively; at point C valve 3 will take over from valve 1. Now phases Y and B’ are conducting in parallel through valves'2 and 3 respectively. At point D valve 4 takes over from valve 2, ete. The difference between the two output wave forms appears across the interphase trans- former, and the output voltage, which is across the positive terminal and the mid-point of the interphase transformer, is made up of the two wave forms as shown by the dotted lines (Figure 2.2). Each valve carries half the current in the d.c. system for 120° in every cycle. ; Figure 2.2. Double star interphase transformer connection (The point vertical line crosses the thick curved line) “A” is the point where the (Cc) BRIDGE CONNECTION (FiGuRE 2.3) In the bridge connection it can be seen that two valves are connected to each phase terminal, the one with the anode connected to it (shown on the upper side of the bridge) and the other with the cathode connected to it (shown on the lower side of the bridge). Thus when the current is flowing out of the phase winding it flows through the upper valve. | Tilia sea aii ae i te nl at A a DIFFERENT TYPES OF VALVE ARRANGEMENTS 13 This occurs when the phase voltage is positive, that is when the anode voltage of the upper valve is positive with respect to its cathode. Similarly, when the current is flowing into the phase winding it flows through the lower Y'6 B'? R'4 Y'6 B’2 a i i aii ia i i id 586 68/ 1&2 223 324 485 526 68! /2? 283 Figure 2.3 Bridge connection eer e eee remanence COR ae Cai 14 TYPES OF CONVERTER CIRCUITS AND VALVE CONNECTIONS valve, and this occurs when the lower valve is negative with Tespect to its anode. Starting at point A, phases R and Yare (as shown in Figure 2.3 [a]). This state valve current. TABLE 3 Comparison of different types of valve arrangements Connections Parameters 3-phase Double 6-phase bridge star diametral 0.427 Va 0.855 Va 0.741 Va 1.045 Va 2.09 Va 2.09 Va 1. Transformer secondary voltage (r.m.s.) Peak inverse voltage Average valve current 0.33 Ia 0.167 Ja 0.167/a Transformer secondary rating .. 1.047 Py 1.481 Pa 1.814 Pa 1.047 Pa 1.047 Pa 1.283 Pa 0.0404 0.0404 0.0404 2. 3 4, 5. Transformer primary rating 6. Ripple factor 7 Peak valve current Ta 0.5 Ia Ta Comparing the bridge conn former secondary utilis as that of the former. voltage though they car. stresses are more troubl. ection with the interphase double star connection, the trans- ation of the latter is poor, though the primary rating is the same The valves in the latter have to withstand twice the peak inverse ry half the current as compared to the former. esoie, bridge connection is preferred. Since the voltage phase voltage is negative, that is the cathode voltage of the TRANSFORMER CONNECTIONS 15 = = aD = a Bridge connection Double - star connection Diametral connection Figure 2.4 Voltage across the valves Figure 2.4 gives, to scale, the voltage stresses on valves in each case for the same d.c. output voltage. Equally important are the sudden voltage changes during commutation, which create oscillations and produce much more stress on the valves. These considerations show the bridge connection to be the most practical arrangement. Since the cathodes of the valves in a bridge connection are at different voltages it is necessary to use single anode valves: but this is no disadvantage, considering the power requirements, since the multi-anode tank will be too bulky and single anode valves will be preferable. In addition, such high voltage levels as are now required can be achieved only by single anode valves. - The term single anode valve strictly means a single phase valve and does not exclude the possibility of using several parallel-connected anodes in the same tank in order to mect the demands of heavy current. i The best feature of the bridge circuit is the ease with which are-backs can be dealt with by blocking all valves by making the grids negative. It is expected in most cases that after a back-fire the valve loses its grid control, and continues to conduct in both directions until the current is interrupted, and some time must be allowed for deionisation. In diametral and double star connections the only effect of blocking the grid is to prevent further short circuiting of two phases, but alternating current still flows in the phase through the faulty valve, the load, and the neutral. The only remedy for this is to trip the a.c. i circuit breaker. In the bridge connection there is no possibility of any such occurrence once the grid is made negative. Two valves conduct in series with the load all the time, but even if one valve is faulty the other valves are all in good condition to block the circuit. : In diamcetral and double-star connections, when there are units connected in parallel, each unit has to be provided with a separate inverse current circuit breaker on the posi- tive side or in series with cach anode to trip when arc-back occurs (to restrict the short circuit away from the rest of the units in parallel to the back-firing valve). In the bridge connection this difficulty docs not arise due to the same reason given above, and several units can thus be connected in parallel without individual d.c. circuit breakers. For a bridge connected valve set, the transformer may be connected in any one of four ways: (a) Delta/star (6) “Delta/delta (c) Star/star (d) Star/delta. 2.3. Transformer connections i ' petnladtennttae tls sini tLe AT en on anne mest naroananetae ance nectar otc nnn annem nes aca” has TYPES OF CONVERTER CIRCUITS AND VALVE CONNECTIONS 2.4 Interconnection of bridge units all t . [ 1 i | { { L : 7 : . 7 To increase the capacity, bridge units can be connected together in a number of ways: | (a) By earthing the bottom bridge unit and connecting all bridge units in series. | (6) Earthing the centre point, the units on cither side giving positive and negative voltages with respect to earth. (c) Connecting in parallel through an interphase transformer. Figure 2.5 Main ways of interconnecting bridge units Figure 2.5 shows these three types. When two bridges are connected in series or parallel two different transformer connections should be adopted, so that the two units are displaced by 30° and both work together as a twelve-phase system thereby reducing the harmonics. Such an arrangement is shown in Figure 2.6. For a transformer with a star-connected primary and a secondary-to-primary turns ratio of Ny (=np/m): E,,=E,/V3 and E,=E, . Ny/\/3 For a transformer with a delta-connected primary and a turns ratio of No (=n! /ny'): E,,'=E, and E,’=E, . Nz | But E,=E,’ since both bridge units are identical: | E, . Ni/\/3=E, . Ne | M/Ne=V/3 Thus, (12/11) . (y'/ng')=/3. i i (a) If the secondary turns of both transformers are equal, i.e., Ng=ny’, then ny'=+/3. ny. a} Thus the number of turns on the delta primary winding will be 1/3 times the number of turns on the star-connected primary winding, and the current in the delta primary will be 1/x/3 of that in the star primary. There is a consequent reduction in the copper cross- section of the conductors in the delta primary which need only be 1/1/3 in area of that in the star primary. (0) If the primary turns of both transformers are equal, i.e., m=’, then Ng=V/3 . ne’. Thus the nuraber of turns on the secondary of the star-connected primary transformer BMH) a) f om | | INTERCONNECTION OF BRIDGE UNITS 17 7 eee! Figure 2.6 Two bridge units connected in series, to form a 12-phase rectifier must be /3 times the number of turns on the secondary of the delta-connected primary transformer. But since the number of primary turns in both transformers are the same, the flux in the delta-connected transformer will be 4/3 times that in the other transformer | core, and this will also be the ratio of core areas. Four units displaced by 15° from each other will work as a 24-phase system, and so on. = Such a phase shift, however, requires the provision of phase-shifting transformers and hence involves considerable extra expense. It is considered unnecessary in practice to increase the number of phases of a system to more than twelve, since reactors are invariably pro- vided on the d.c. side for other reasons and the reactive power equipment on the a.c. side can be suitably modified to absorb a large proportion of the harmonics. Harmonic pheno- mena are discussed in detail in Chapter 10. An advantage of increasing the number of bridges is to increase the reliability of the transmission system. The larger the number of bridges, the smaller will be the proportionate reduction of power transmission capacity if one bridge fails; when the system is not operat- ing initially at full load, the power can be maintained by increasing the current. Further- more, a large number of bridges in series enables the individual bridge transformers to be designed for a reasonable overvoltage; by this means, if a bridge fails and is out of com- mission for some time, full power could be maintained by increasing proportionately the operating voltage of each of the remaining bridges. It must be mentioned here that when a bridge is faulty, its by-pass valve (the 7th valve in each of the bridges shown in Figure 2.7) opens, so that the remaining bridges can still continue to operate (Chapter 6). When a bridge fails suddenly, it is accompanied by transient over-current and voltage, and the larger the number of bridges, the less will be the effect of these transients, shady a cE 18 TYPES OF CONVERTER CIRCUITS AND VALVE CONNECTIONS The bridge number cannot, however, be increased arbitrarily, since each bridge requires the provision of its own auxiliarics and main transformer. At some stage it is necessary to consider increasing the power level by series and/or parallel connection of valves, The connecting of bridges in parallel raises difficulties in ensuring equal current distribu- tion through the medium of an inter-phase transformer, and is not favoured. Fortunately, \_the manufacture of valves of high current ratings is practicable. The possibility of using one transformer, with two secondaries and one primary to connect two bridge units is attractive, and such an arrangement is shown in Figure 2.7. The two secondary windings of any one transformer may have different connections, i.e., the one in star and the other in delta.2”) The most important advantage of two secondaries and one primary is economic. A further merit is that when a fault caused by a back-fire occurs on any one of the secondaries, the short circuit current in the primary will be due to one secondary fault; as the primary is designed for the current of two secondaries, the tendency will be to produce less severe stresses on the transformer; this depends, however, on the precise winding arrangement. There are, however, disadvantages with such a scheme. If, due to a back-fire, there is a short circuit in one secondary it will result in a fall of voltage and consequent fall of magnetic flux in the core of the transformer. This will reduce the voltage induced in the second secondary, but due to the self-capacitance, mutual capacitance, and capacitance with respect to carth, this sudden change in flux causes high frequency oscillation in all windings. This may cause a breakdown of any valve in the other secondary, and may thus result in a double short circuit and a much more severe fault condition. The method of avoiding such difficulties is to provide a high degree of screening, a surge divertor across each winding, and adequate damping circuits. This protective equipment is necessary also to take care of the oscillation produced by the commutation process (Chapter 6). This method of using two secondaries is practicable, and is being adopted in the Soviet Union for the Stalingrad to Donbass h.v.d.c. project.“°) Adequate damping arrange- ments and surge divertors are being provided. Reference may also be made to the Troll- hattan h.v.d.c. laboratory in Sweden, in which this type of transformer was provided.( It was not, however, adopted for the Gotland Scheme. It should be noted that interaction betiveen windings on the same core can be compensated by the use of special winding arrangements; this topic is dealt with in Chapter 3, Scction 3.7.5. : 2.5 Valve arrangement withia a bridge The advantage of connecting valves in series, as distinct from bridges in scries, is that the increase in converter rating is obtained by the provision of one large capacity transformer rather than a number of small ones. Connecting valves in series is not, however, a very straightforward matter. Each valve cathode, together with its auxiliaries, has a large capacitance to earth whereas between adjacent valves, the capacitance is small. The result- ing voltage distribution is uneven with the highest voltage occurring across the valve with the highest potential with respect to earth. A better voltage distribution can be obtained by the use of potential dividers across the series arrangement of valves. In theory the dividers may be purely capacitive but in practice arn erent nel eevee ne emit iinsen naan dei oes mdr 19 UO pJBAUE HOI] 99 Ac- Ag- | : wae © is Sayoy7 Weed, sala +) rt Holl COG, © |. wossiisuoy Ot ; At At [I \ ) vu r exe Q fi 7 Woe & | _eoro d TO : ary AZt ae < ! VALVE ARRANGEMENT WITHIN A BRIDGE oa p : _ Figure 2.7 ‘Laboratory h.v.d.c. system with two bridges supplied by each transformer ‘Uap ung §=s0sow y Et =——© n> a q | { | is 50) Te ee ea OE NN eae eee 20 TYPES OF CONVERTER CIRCUITS AND VALVE CONNECTIONS combined resistance-capacitance dividers are desirable. Purely capacitive dividers are perfectly satisfactory under system steady-state and transient conditions, and do not con- tribute to loss of real power. They contribute, however, to the parasitic oscillations which accompany the sudden voltage changes ensuing from starting and stopping of the valves (Chapter 6). On the other hand, a resistive chain materially assists in damping out such parasitic oscillations, but contributes a small additional loss of power and provides a poor voltage distribution under transient system conditions. Ina system using a purely resistive divider, a suddenly applied voltage results in the larger proportion appearing across the valve most remote from earth, the final voltage distribution only being attained after a time dependent on the time constant of the resistance elements and the stray capaci- tance.6% | These considerations lead to the use of resistance-capacitance dividers. Clearly as the valve number is increased, the natural voltage distribution becomes more uneven, and these dividers become more elaborate and hence more expensive. Hence in view of the large stray earth capacitance of the valve cathode and its auxiliaries, it is impracticable ever to consider more than, say, four or five valves in series. Some experiments have suggested that by increasing the number of valves in series to two or more, arc-backs can almost be eliminated since the possibility of all valves in series backfiring simultaneously is very remote. The attainment of engineering reliability by such means is open to question, however. With high-voltage valve chains partial breakdown can occur in which one valve, due to excessive ionisation, can conduct a small, but sufficient, current through its voltage divider circuit to act as a short circuit; the whole of the back voltage would then appear across the remaining valves, which would tend to fail one after the other until complete backfire resulted. Series connection of valves may reduce the incidence of “arcing-through”, ice., the failure of grid control. If one valve of a chain fails forthis cause, the other valves will pre- vent the series combination from conducting. At the same time, the chances of ignition failure of the series combination may increase if one of the valves fails to fire. These faults have adverse effects on invertor operation and cause system short circuits (Chapter 6). The advantages of valves in series finally depend upon the frequency of partial breakdown in a series combination, compared with the likelihood of breakdown of one valve alone in a bridge arm. Clearly, with two valves in series, failure of one valve will put the whole of the voltage across the other; if partial failures are not rare, then each valve must be built to withstand full voltage and the rated voltage of a bridge unit cannot be increased beyond what would have been the case with only one valve per arm, although a great reliability would be achieved. _ Advantage in this respect thus lies with three or more valves in series, each valve being designed fora proportionate share of over-voltage in the event of one of them failing. If parallel connection of valves is adopted, the incidence of shut-down due to ignition failure is reduced, since the companion valve will continue to conduct although it will be overloaded. This is probably well within the capability of a valve for a period of a few cycles since it will have been designed for short-time overloads. This purpose may alter- natively be served. by providing two or more, parallel-connected anode structures with a common tank and cathode. It should be noted that valve development in Sweden indicates SYSTEM ARRANGEMENTS ON D.C. SIDE 21 that a valve of any current capacity can be made economically by increasing the number of anode structures in parallel. Increasing the number of valves per bridge means decreasing the number of bridges and hence diminishing the advantages gained from using bridges-in-series. A distinction must clearly be drawn between valves-in-series and bridges-in-series, at this stage. Valves-in- series increase the reliability of the bridge into which they are connected and hence the reliability of the transmission system. Bridges-in-series, however, increase the reliability of the transmission system in the event of failure of a bridge, and in the same event decrease the transient over-voltages and currents associated with the remaining bridges, Thus the greatest reliability of converter plant is only obtained by compromise. Since each valve needs its own auxiliaries, and each bridge unit its own transformer (or at least its own secondary winding), the cost of a converter station rises considerably as each of these items is multiplied. Arbitrary increase in the number of bridges and/or valves-in- series is thus out of the question, and the total number of valves is likely to be dictated largely by the maximum size of valve available. At the same time, it is very desirable to provide a minimum of two bridges in series, one of each polarity with respect to earth, irrespective of the maximum valve size available. In the design of the Stalingrad-Donbass transmission scheme,“ eight bridges have been provided for 800 kV (£400 kV to earth), and each arm of a bridge unit consists of two valves in series. In normal service, the valves will be operated on half their rated voltage and only during the comparatively rare occurrence of partial breakdown and backfire will the healthy valve of an arm have to withstand the total voltage stress. 2.6 System arrangements on the d.c. side The methods of arranging bridges and valves discussed above give rise to three main types of d.c. transmission system, as shown in Figure 2.8. 2.6.1 Single conductor system with ground return [Figure 2.8 (a)] In this arrangement earth is used as an active conductor and has the advantages of low capital as well as low transmission costs. The earth return, however, has certain disadvantages: (i) Interference with railway signalling (Chapter 9). (ii) Corrosion of cable sheaths, pipes and any other metallic equipment in contact with earth (Chapter 9). q (iii) In the case of the sea providing the return path, there is the problem of magnetic compass error (Chapter 9). (iv) Interference of system harmonics with communication circuits and channels (Chapter 10). Of these (i) and (ii) are fortunately caused only in the neighbourhood of the electrodes, whilst (iii) and (iv) are caused along the whole of the power transmission route in the ncighbourhood of the conductor. The problems of (i) and (ii) arise from leakage of the earth current into adjacent equip- mentincontact withearth. Further away from the electrodes, the current passes into deeper 2th ; ; (d) = 4 —_— Vy/? (Cc) Vd/2 t Figure 2.8 Main types of system arrangements viewed from the d.c. side layers of the earth’s crust and hence causes no disturbance; (iii) is caused by the magnetic field set up by the direct current in the conductor; (iv) arises from both capacitive and magnetic coupling between the power conductor and communication circuits, ot 2.6.2. Two conductor system without earth connection [Figure 2.8 (6)] The transmission of current is now constrained to the two conductors. Problems (i) and (ii) above (Section 2.6.1.) are absent and (iii) and (iv) ate greatly reduced. This method is expensive, however, and undesirable from the point of view of protection. 2.6.3 Two conductor system with mid-point earthed [Figure 2.8 (c)] This is the most advantageous and favourable arrangement, It may be regarded as a duplication of the system in Section 2.6.1. [Figure 2.8 (a).] The problems of (i) and (ii) are eliminated and (iii) is greatly reduced; (iv) may still be present, if there is phase displace- ment between the alternating components in the output on each side of the earth connec- tion, in which cast some harmonic components will return through the earth (Chapter 10). ca co (4a mm nm | | SYSTEM ARRANGEMENTS ON D.C. SIDE 23 Reliability is increased, since the system is equivalent to that of a double-circuit system. In the event of a d.c. line fault, the unfaulted side can continue to supply power through the earth as an emergency return path. This does not materially effect the problem of corrosion since this is a long term effect. The problem of compass error depends on the attitude of the maritime authorities and is in any case subject to agreement. The problems of inter- ference with communication and railway signalling systems and apparatus demands con- sideration, although their solution is possible without very serious difficulty. One further possibility is that of two or more, mid-point earth circuits being operated in parallel, thus increasing still further the overall reliability of the transmission system. But in view of the high level of reliability attainable with two or more valves in series per arm and multi-bridge operation, double systems of this sort are unlikely to be used unless there is the possibility of converting a double-circuit, a.c. transmission system to three, mid-point earthed, d.c. systems operating in parallel. cm et ee Et CHAPTER 3 Bridge Rectifier and Invertor Parameters References: (31) to (37) inclusive, (43). 3.1 Assumptions made to simplify the calculations In order to simplify the calculations, it is convenient to make the following assumptions: (a) The load on the d.c. side has infinite inductance, so that the output d.c. is constant, (6) A.C. busbar impedence is zero, and thus the a.c. system has infinite capacity; in other words, no account has been taken of the effects of converter load on the a.c. voltage. This assumption is quite valid for practical purposes, when the converting plant capacity is not large compared to the system capacity, but when this is not the case it involves some error. When the effect of converter load on a.c. voltage avail- able at the a.c. busbar of the converter is taken into account, the calculations are complicated for practical purposes. Such a theoretical analysis has been made by Uhimann, 6) (c) Arc drop is neglected. This can be taken into account, later since are drop is practically constant for all loads. In high voltage valves are drop is negligible com- pared to the output voltage. (d) The magnetisation current and the resistance of transformers is neglected. Only an approximate and elementary analysis of rectifiers and invertors is to be considered in this chapter, for purposes of general background and to provide a basis for subsequent development of the subject. 3.2. Rectifier parameters It is convenient to carry out these calculations in four stages. 3.2.1 Ideal conditions: zero transformer winding reactance and no grid control Figure 3.1 gives the d.c. output voltage and current in each valve and phase, assuming a star connected secondary. (i) D.€. output voltage. The average value of the output voltage V, can be calculated easily by dividing the wave shape into symmetrical parts and then finding the average of one part, say ABCD. Since the voltage between phases at point X is a maximum, the equation of the instantaneous output voltage will be: e=/2 Ecos wt, where E£ is the r.m.s. secondary voltage between phases. Tw TG Yorn V2 Ecos wt. dut 2 6 regmemenneregeremteytaeen nee cap 2 37 aoe > Mt RECTIFIER PARAMETERS 25 LAC. a t voltage ~ Z 4 6 2 Figure 3.1 D.C. output voltage and current in each valve and phase; ideal conditions, no grid control = A CT CTR ee a re ET EOE neat 26 BRIDGE RECTIFIER AND INVERTOR PARAMETERS T +2 6 bi z[ sin wr | (ii) RMS. value of secondary current. At any instant two of the six valves carry current. The secondary current alternates between limits of +I, and —J, over a period of 120° in cach half cycle; hence there is no d.c, component in the transformer secondary. current. R.M.S. value of the 1 {7 transformer secondary current, =A/ a i? . dust -4/ f ae Beals suck (3.2) 3 Transformer rating= 1/3 El 7 V2 Tress Vox I, =F Mola =1.047 V, [,=1.047 P, 3.2.2 A finite transformer winding reactance and no grid control It has been assumed in the above calculations that the current transfer from 1 to 3, 3 to 5 and so on takes place instantaneously as soon as the anode voltage of the latter becomes more positive than the former. But actually the current transfer takes some finite time, since the current change from J, to zero and zero to I, in zero time involves infinite rate of change of current, which is impossible if the leakage inductance of windings is taken into account. The current change takes place in some finite time as shown in Figure 3.2. During commutation from 1 to 3 the current in 3 (phase Y) rises from zero to T, and the currentin 1 (phase R) falls from I, to zero. During this period both phases and valves conduct simultaneously and this period of commutation is also called angle of overlap. It is clear from Figure 3.2 that since phase R is conducting through valve 1 and since phase Y starts conducting through valve 3 it short circuits phases Rand Y, starting from point A. The voltage which will circulate the short circuit current between these phases in direction Y—3—1—R is, e= V2.E.sin wt. le ete Im ‘2 ~ z ) ~ 7 g 2 & J (™ - So Sha we a ae hes RECTIFIER PARAMETERS 27 /R a 58 Figure 3.2 D.C. voltage and current; finite commutation re- actance, no grid control Neglecting the resistance of the windings and the are drop, which is very small compared to the leakage inductance L, of each phase of the winding, the short circuit current is given by: di 2L ave E sin wt Integrating, pene te E (=2")+¢ 2L w when t=9, i,=o. E haart ya E V2 wL Tn fact this short circuit current increases to J, in valve 3 and reduces the current in valve 1 to zero and stops, since valve 1 cannot conduct in the opposite direction. when wt=y,, i,=J,, (1—cos y,) .......ceceeeeeees (3.3) E V2 wh During commutation the output voltage follows ab, the mean of the voltage between Rand Yas shown, since there is equal drop in the two windings. Thus there is some drop in the output d.c. voltage by an area 8A (abc in the figure) for every 60°, the average value of which will be 8V, given by: and i,= (1—cos wf) oh 28 BRIDGE RECTIFIER AND INVERTOR PARAMETERS Ww 6A/==8V af; 8A=3| 7° 4/2. Esin wt. det =} 2. El—cos wt]” ; ° v7) E (1—cos y,) 3 5V= —_ TOOS Va) wee c ccc cecccnvcnccccecs 4 it * Fag Fa cos ¥,) ., (3.4) ' 32 E (1 ft From cquations (3.1) and (3.4)V = v,—-sva3¥? ( ses %0) i 7 V. = > (1-++cos y,) From (3.3) and (3.4), sV=I,. on Sone ee eee eee e cece teen ec ec eee cacene (3.6) wT Thus the voltage drop is Proportional to J, and the effect of commutation appears as a resistance on the d.c. side, equal to (3wL)/z which results in a voltage drop of 30L/n . I, in the d.c. output voltage. But it does not mean that there is any loss of power, since it is not actually a resistance. Since the magnitudes of the a.c. current and voltage are the same, the reduction in d.c. power from T,.V, toI,.V, (1+cos y,)/2 is caused by the introduc- tion of lagging power factor on a.c. side, resulting in reduction of power from 1/3 E, I, to V3 E, I, cos 4, where £, and J, are the line voltage and current, respectively, on the primary side of the rectifier transformer. “A= C08 $® (14008 ¥,)/2 0. eee cece ececc eee. (3.7) This is an approximate value of the power factor, since the introduction of the angle y changes the r.m.s. value of the current on the a.c. side. nN we un ¥% 3.2.3 Grid control, transformer winding reactance neglected The d.c. voltage output can be controlled from V, to zero by delaying the firing of valves as shown by Figure 3.3 Now the valve 3 will take over current from valve | at b, instead ; of a. The calculation will be exactly the same (—2/6+2) to (7/6+<). as in equation (3.1) except that the limits will be + 7/6 +4 V, ue [v2 Ecos wt dwt as — 7/6 +a 3/2 =—— E(2sin 7/6 . cos a) wT 32 =—— Ecos a=V, cosa..................., (3.8) wT atta ta een Gar os ee ee em oo B oN fr oS a xo ze 53 a ws xT ; GARG RECTIFIER PARAMETERS 29 Figure 3.3 D.C. output voltage; grid control, no commutation reactance 3Y 58 Figure 3.4 D.C. voltage and current; finite commutation re- actance and grid control xX Y= yet 4 5 For zero delay cos a=1, and the output voltage is a maximum, V,. 3.2.4 Grid control and transformer winding reactance taken into account (Figure 3.4) The equation of short circuit current in Section 4.2.2 is: di . 2L A /2 Esin wt. . ._ V2 E (cos wt) t= op Fe ay SEER / 30 BRIDGE RECTIFIER AND INVERTOR PARAMETERS ; when wf=a, i,=0 E C= Val’ cos a i = (cos a—cos wf) = a— w V2 wh when wt=a--y, i=l, "= 7. [cos a—cos(a+y)]}.....0.0.....000., (3.9) aty As in equation (3.4), 84 =H V2.£E.sin wt . dwt a oe [cos cos (a+y)] V2 a at+y 3E Pw a [cos a—cos (a+y)] Vv, =z [cos a—cos Coy EEL (3.10) Vi=V, cos a—8V V. =a [cos a+cos (a+y)}.................... .11l) As in equation (3.7), the approximate power factor on the a.c. side: A = cos $=} [cos atcos Cea y)h feat EE el Uy) (3.12) - 3 woL From (3.9) and (3.10), 6V=I,. = (453-6) jee Te LL | (3.13) Thus for the same current, the voltage drop due to commutation is independent of the delay angle. This means that the commutation area 84 is always the same for the same current. When the delay angle is increased, the angle of commutation will decrease, since the commutating voltage (vertical distance between phase voltages) increases. Thus the greater the angle a, the less the angle y for the same current. ; The equivalent circuit of the rectifier will be as shown in Figure 3.5. Calculation of the r.m.s. secondary current, taking a and y into account, will be dealt with in Section 3.9 3.3. Rectifier characteristics : E (a) When a=0, = OL (cos yD EU EAL (3.3) I is finite {co: cos (a+y)] (3.9) n in: = Ss a— BTV] cece cece cece cece cere ences le when a 1 ‘d V2 OL Sa : | *: For the same current, (1—cos ¥)=[cos a—cos (@+7)] 0.2.00 (3.14) { \ zalt << OPERATION OF AN INVERTOR 31 From equation (3.14) a set of curves of « against ycan be plotted for fixed values of Yo), as shown in Figure 3.6. . The angle shown on the Y axis represents y, and y. Thus when E and wl are known, y, can be calculated for any value of d.c. for which the angle of overlap is to be found. The curve corresponding to each particular value of y, will give the angle of overlap y for any delay angle a. 7 For example, suppose that E//2 wL is known and is constant (E is constant), then from equation (3.3) y, may be calculated for any value of current Jj. Now suppose that y, for some value of J, is 18.5°, then by inspecting the curve shown by the dotted line (Figure 3.6), which starts from y=18.5° and a=0, it is possible to find the angle y for any angle a. Thus for angle a=58°, y will be 3.5°. Again for another value of J,, the corresponding angle y, may be calculated and the angle y found for any angle a from the corresponding curve. For particular rectifier arrangements, the curves of y, against J, for different values of E may also be plotted. (b) Another method of plotting these characteristics is as follows 6) Tiny [cos a+cos (a+y)]...... ccc eee ee eee eee (3.11) Ty. vere [cos a—cos (aty)] ... eee eee cece cece e eee eee (3.9) Thus from equations (3.9) and (3.11), different values of V,/V, can be found for different values of y for fixed values of a, and then plotted against /,. \/2 wL/E. Similarly another set of curves of V,/V, against I, . »/2 wL/E can be plotted for fixed values of y. These curves are shown on the upper portion of Figure 3.7. It is evident that when E and L for a rectifier are known, the relation between V,, y, a and I, are easily obtained over the whole range of operation. . 3.4 Operation of an invertor Consider first a bridge unit working as a rectifier. V,=V, cos a (neglecting commutation angle) ..........-+-e sees eee eee (3.8) Figure 3.8 gives the output voltage for varying control angle; (a) is for the normal rectifier with a small delay a and (6) is for the case when a is increased beyond 60° and it is seen that there is some negative voltage. If the load on the rectifier was resistive, its operation would be intermittent since con- duction is impossible in the negative direction. - But when a large reactor is provided, the difference between the negative and positive voltage areas gives a resultant voltage output on the dic. side. : At 90° delay-{c) the positive and negative voltage areas become equal and the average Voltage is zero. At 180° delay (d) the negative area becomes as much as that for the rectifier with no delay. : : V,=V, cos 180°=—V, Now if an external d.c. voltage is applied which forces the current and overcomes this negative area, the current will flow from anode to cathode in opposition to the induced “ 7 i; — > neiraecemrnts diate bembaleerel a peg pe -s rome L t t 32 BRIDGE RECTIFIER AND INVERTOR PARAMETERS Sal 7T ly : laa | Voss -VieV, cos -3wl.T) l (eae Figure 3.5 Rectifier characteristic equivalent circuit : NTT 50 Ho \ ! iS OA 8-5 PVE TY < AC LENE NS \ | a 8 20 : eee LAT |\ Rectification fnaversion A [= = {0 a ] LE] Rye || i he o /3\\\\ 1 4 } & RAACSCECCEC OA ATI oe ONIN NSP f Ss \ |_| - = holes =o SSS a = t DB t eee ee c oe = 0 20 40 60 60 100 120 40 160 10 Angle of grid delay ce ¢ degrees) .. Figure 3.6 Complete rectifier and invertor characteristics (The change-over ffém rectification to inversion will occur at the points where the output voltage is zero that is where the angles a and y are such that a + y/2 = 90°) : “a OPERATION OF AN INVERTOR 33 a /0 Vd Vo 0-5, o g eeoe & =40° SS ~ S =50 & JyV2uml. Ty bh Ee h5P* 0 TAL S: 50 g =e ‘ = g 60 S S a Figure 3.7 Complete rectificr and invertor characteristics e.m.f. in the transformer secondaries, thus indicating that power is being supplied to the a.c. system. The rectifier therefore becomes an invertor. The conversion of electrical energy follows the principle of the transformer. Ifthe current Se I = 34 Negalhive areas Positive areas /R BRIDGE RECTIFIER AND INVERTOR PARAMETERS 2yY 38 @ Figure 3.8 Operation of an invertor © (© Figure 3.9 Invertor output voltage and current ene ecenipensa eee ee amin arena: oe i Sate i a nna ee ila tele IL asclcaai INVERTOR PARAMETERS AND CHARACTERISTICS : 35 ‘ flows in the direction of induced e.m.f. it means a release of energy, i.e., rectification, and when current flows in opposition to the induced e.m.f., it means acceptance of energy, i.c., inversion. The commutation cannot be delayed beyond S (Figure 3.9) since after this point the voltage of phase R becomes negative with respect to that of B. In fact some angle y has to be left for completion of commutation from valve 5 to valve 1 (commutation angle) and also some angle §, to enable valve 5 (which has stopped conducting) to deionise; otherwise valve 5 will fire again and take over conduction from valve 1, Figure 3.9. Thus valve 1 will fire at point P, an angle 8 before point S. This angle must not be less than y+,. The period 8, is called deionisation time. The current J, in 5 starts falling and the current in 1 starts rising at P, and commutation ends after angle y from P. After an angle 5, from this point, the deionisation is complete in valve 5 and its grid regains control and the valve is blocked, so that it does not take back current from 1 after S. This angle, 5,, may be of the order of 10°. Thus valve 1 is prevented from firing up to point P by the negative bias on its grid, even though its anode is positive with respect to its cathode. This means that grid control is essential for invertor operation. For invertor operation the current will always lead the voltage, due to the minimum angle y+ 8, and hence the invertor should be supplied with some reactive power, either from the a.c. system or by connecting a synchronous or static capacitor on the a.c. side of the invertor. The availability of an a.c. supply is an ‘ essential requirement for an invertor. 3.5 Inyertor parameters and characteristics The calculation of the d.c. voltage, the commutation angle, etc., can be obtained in exactly the same manner as with a rectifier. : As in equation BB), Fel CON Bs al cells snes sala e neld p belee bia (3.15) where V,=3/2/7 E (neglecting commutation angle) cc _- — As in equation (3.10), v2 (cos 5—cos B).........-.0 eee hobbies salencpicle tld (3.16) In the case of invertors 5V will increase the back e.m.f. V,, as can be seen from Figure 3.9. j : ie., V,=V, cos B+ 8V =V, cos p+ (cos &—cos f) a V. == (cos f+c0s 8) . LE dn G.17) E i ion (3.9), = B—COS) oe eee eee eee ee wees i in equation (3.9), Ly V2 ok (cos 5—cos f) (3.18) From equations (3.17) and (3.18), Vz=V, cos B+I,. BUTE L (3.19) Tv Also V;=V, cos 8—8V V. =a) (Cos B+COS 8)... cece cece cence eeeee - (3.17) L i Bc || 1 Oe enna pe meen eT eT er rane meron = ‘ie { da ta A lamin en 36 BRIDGE RECTIFIER AND INVERTOR PARAMETERS 3 wL HV 608 By cece cece ee (3.20) wv 3 al Wp ede delle Me ele) (3.21) 7 3 wL/7 is the equivalent resistance for invertor commutation. The equivalent diagrams for an invertor are shown in Figure 3.10. Figures 3.10 (a) and 3.10 (4) represent the situation when the angles 8 and 5 respectively are maintained constant. In the latter case, it can be seen that the invertor voltage decreases as the current increases, Bul I +3uL +i dc. vollage from Te LL rechifien E = Vi cosB Y, cos B+ Ly Sul. iF UT naif @ — Gwl) T { dc. voltage From rectifren YX cosd- Id 3wl. (6) Figure 3.10 Invertor equivalent circuits Invertor equations can also be obtained directly from the rectifier equations by putting a= a—B and y=8—8. The rectifier curves can be extended to invertor curves for values of a>90° as shown in the right-hand side of Figure 3.6 In the figure, the “commutation limit line” is drawn, beyond which successful commu- tation is not possible. Thus, if y=15°, the angle a cannot exceed 180°—15°=165°, as indicated by the line. This is without taking account of the deionisation angle, which may be approximately 10°. Thus a line parallel to this commutation limit line may be drawn which will take into account the deionisation angle 5,. = ENE ni raenneN ahay SOA? -geronnemeenarer ert peyote ap vend topecw gnome + ae a ARR UE urrent COMMUTATING REACTANCE . 37 The invertor characteristics can also be plotted from the equations: mt (Cos B+C0S 8)... cece cece ces ccccsvccce (3.17) ° V2 ol and J,. = (C0S| 6 C03 ED (3.18) in exactly the same manner as for a rectifier, as shown in the bottom half of Figure 3.7. The approximate power factor on the a.c. side will be (cos 8-++-cos Byer (3.22) and will be leading. It should be noted that the calculations above represent the rectifier and invertor con- ditions for a commutation angle less than 60°; if this angle is exceeded, the operation is — no longer represented by the same expressions since two commutations will take place simultaneously. Since this situation, is unlikely in practice, it is not considered in detail in this book, although the characteristics of Figure 3.7 have been extended to cases of y greater than 60°. It should also be noted that operation for the case of « less than 30° and y greater than 60°, simultaneously, is not possible; if for a=O, the current is so in- creased that the corresponding angle y is greater than 60°, then the firing of the valves will not take place at a=0 but will be automatically delayed so that y=60°. This situation continucs until a=30° and then y can be increased; under such conditions, the equivalent commutation resistance becomes 9wL/m, three times the normal value of 3wL/m as given by equation (3.6). 3.6 Commutating reactance In the previous discussion the commutation reactance of the converter has been taken as the Icakage reactance of the windings only; but in actual installations the commutating reactance consists of the reactance of the leads (which is negligible), the transformer react- ance and the a.c. system reactance. The latter must be considered, irrespective of whether or not the r.m.s, value of the a.c. busbar voltage is assumed constant (under the assumption of zero busbar impedance). The commutating reactance of the transformer windings is the leakage reactance be- _ tween the secondary windings and their respective primary windings, and can be calcu- | lated or measured. The commutating reactance of the a.c. system usually varies in some degree according to the system load conditions, and the approximate normal value can be found by measuring at the particular point where the converter is to be connected. 3.7 Valve voltage in bridge connected invertors) 3.7.1 Valve voltage in a single bridge connected invertor As stated before, the process of commutation is a short circuit between two phases. | The voltage between these two phases disappears and drops across the commutating | reactance X¢, which consists of the leakage reactance of transformer X, and the system { reactance Vs [Figure 3.11(@)]. In a three-phase system, during the disappearance of one Voltage, the other two voltages adjust themselves to balance in such a way that their com- ponents disappear in the direction of the short-circuited voltages. Figure 3.11 Commutation in a single bridge invertor £3! (Q) In Figure 3.11 (6), E,, E, and E3 are the voltages between phases R and Y, Y and B, and B and R respectively: the voltages E,, Es change to E,’ and E3' when E, dis due to a short circuit between phases R and Y, as shown. This condition produces a phase displacement of both voltages, E, and Es, by +30° together with a reduction of the amplitude to 4/3/2. Thus if (due to the short circuic at instant YX’) the voltage ¢; disappeais, Ex increases to es’ and 3 decreases to e,’. The same thing happens during the period of commutation, as can be seen from Figure 3.12 (a). The voltage across valve 1 will be as shown in Figure 3.12 (6). During the conducting period the Voltage across the valve can be assumed to be practically zero, As soon as the commutation is over, the voltage becomes E, (voltage between phases Rand Y), which is negative at first and changes sign after an angle, the minimum permissible value of which is 8,. During commutation from 2 to 4, Es vanishes produces a dent in the voltage across the valve. appears and £; changes accordingly. This At the end of commutation from 3 to 5 the valve voltate changes from E, to E; (voltage between phases R and 8). During 6 no "ure = ses { this * A etna er i A I VALVE VOLTAGE IN BRIDGE CONNECTED INVERTORS 39 commutation from 4 to 6, a dent is formed in the same way by changes in E3. Thus the voltage across the valve 1 will be shown as in Figure 3.12 (6). 3.7.2. Valve voltage when two bridge units operate together displaced by 30 degrees When there are two units together as shown in Figure 3.13 (a), the system reactance, which is part of the commutating reactance, is common to both the units. Therefore when commutation takes place, that is when one voltage disappears, changes occur in the other two voltages of the same unit; that is, disappearance occurs of their components along the vanished voltage. oY 2B AR Figure 3.12 Voltage across any particular valve in a single bridge invertor (valve 1 shown for example). CHET: 40 BRIDGE RECTIFIER AND INVERTOR PARAMETERS Figure 3.13 Commutation in double bridge invertor Besides this, the voltages of the other unit will also change in such a way that their components in the direction of the vanished voltage reduce in the proportion ¥5/(X,+ Xs), since X is the reactance common to both and X, is the individual transformer leakage reactance. ~ porn ~ a ann anc te il a em Nn ee VALVE VOLTAGE IN BRIDGE CONNECTED INVERTORS 41 In Figure 3.13 (6) Ey, E, and E; are voltages between phases of unit J and E’;, E’s and E’; are the corresponding voltages of unit J/ displaced by 30°. If £; disappears Ey and £3 change to E,: and Ey: respectively, so that their components in the direction of E; disappear; E’; and E’schange to E’y and E’3: respectively, so that their components in the direction of E, do not disappear completely but are reduced by a fraction Xs/(X_+Xs5). Es is not affected since it has no component in the direction of E, and is at 90° to it. The voltage across valve 1 will be the same as before; in addition to that there will be dents due to commutation of 2’ to 4’ and 5’ to 1’ as shown in Figure 3.14. The magnitude of these two dents will depend upon the ratio X5/(X,+X;). There will not be any effect eee ee eee eee | ce 4 Figure 3.14 Voltage across a particular valve ina double bridge invertor (valve | shown for example) a | | 42 BRIDGE RECTIFIER AND INVERTOR PARAMETERS Figure 3.16 Compensating reactors in a double bridge invertor 30. Figure 3.15 Effect on commutation angle in a double bridge invertor of commutation of 3’ to 5’ and 4’ to 6’, since during these commutations the commutating voltages E’5 and E’; are at 90° to voltages E, and Ey respectively, as shown in Figure 3.14. Exactly the same analysis applies to two sets in parallel on the d.c. side. 3.7.3. Effect of voltage dents It is the voltage dent A shown in Figure 3.14 (b) which is important, and this feature is magnified in Figure 3.15. It is clear from the diagram that commutation angle y should be less than 30°—58,. Ifthe current is greater, so that the commutation has to be started earlier, it will result in insufficient deionisation time causing commutation failure. When Xs is very small compared to X;, this will be very small but nevertheless ii will still be dangerous. Moreover X; is not controllable and depends upon the a.c. system load and other factors. Thus some method has to be adopted to remove this dent, that is to say, to remove the effect of one unit on the other. 3.7.4 Method of compensation for the effect of one unit on the other This compensation is carried out by introducing compensating reactors Y, equal to 4X, between the primaries of the two transformers and connecting the system terminals to their midpoints, as shown in Figure 3.16. It is quite apparent that during commutation the commutating voltage will drop across X, Xs and X,/4. The common reactance of the two units is Xs. At the same time, because Xp/4= Xz, an equal voltage will be induced in the other half of X, and the dent will disappear. If Xp is less than 4) ‘> the dent will be smaller. If X> is greater than 4X5, the dent will be negative and the commutation margin will increase. A reactor of value Xp, slightly greater than four times the normal system reactance ¥5, may be provided for safety. 7 Ace, ea Lr VALVE VOLTAGE IN BRIDGE CONNECTED INVERTORS 43 7 Primary Secondary Secondary 7777 (4) Ya ; isi Primary ; fqualisi#gg winding Secondary Secondary (0) Figure 3.17 Winding on onc leg of a compensated 2-secondary transformer for a double bridge arrangement ch sii cine 44 BRIDGE RECTIFIER AND INVERTOR PARAMETERS Since the system reactance will decrease with increased load, even a slightly under- compensated Xp will not lead to danger. This method has been tested in Sweden and has also been incorporated in the Gotland scheme.@2 The introduction of Xp increases the commutating reactance X¢ to Xp +Xo+Xp/4, hence increasing the commutation angle, but this can be compensated by minimising the leakage reactance of the transformer. Similar analyses apply to more than two units, and compensation is employed in the same manner. 3.7.5 Compensation in a double bridge arrangement which uses one transformer with two secondary and one primary winding®» ; . The commutation process in one secondary winding produces voltage changes in the i other secondary winding, as discussed above. Compensation may be provided by the use : of two three-phase secondary windings, each of which supplies a bridge unit, on either side : , @ Single bridge rectifier (0) Double bridge rectifier Figure 3.18 Voltage across a valve in a single and double bridge rectifier oa in the the use f side [ [ L [ feath fs (3 RD om VALVE VOLTAGE IN BRIDGE CONNECTED RECTIFIER 45 of the primary winding which is connected to the a.c. system. In this way, an equivalent negative reactance is obtained on the a.c. system side; its value, however, is usually too small and may be increased by division of the primary winding into two parts with space between them. This arrangement of windings is as shown in Figure 3.17 (a), one leg of the transformer only being shown for convenience. Another possible arrangement is shown in Figure 3.17 (6). Each secondary winding is wound around its own part of the divided primary; two equalising windings, connected in parallel, are also provided. _ If it is necessary to increase the reactance between the two secondary windings then an extra reactor can be connected in the circuit of the parallel connected equalising windings. A further possibility is to connect the equalising winding in series with the primary and so use it as a regulating winding. Compensation may also be carried out on the valve side of the apparatus or on other transformers inserted in the invertor plant. 3.8 Valve voltage in the bridge connected rectifier The rectifier valve voltage will be just the opposite of the invertor valve voltage, as can be seen from Figure 3.18. The voltage dents will appear in the same manner as in the case of the invertor, and need not be discussed separately. Two extra dents produced by the working of two bridge units, displaced by 30°, will not cause any harm since they do not lead to any failure and hence need no compensation. 3.9 R.m.s. value of the transformer secondary current The r.m.s. value of the secondary current, when commutation and delay angle are both taken into account, may be calculated by dividing the current waveshape into three parts as shown in Figure 3.19 (a): [cos a—cos (wt-+a)] [cos a—cos (a+y)]’ instant of commencement of the current rise as the zero axis. =I, from equation 3.9, but taking the i,=T,, the Steady-state portion of the current wave. . I, [l—{cosa—cos(wt-+-a)} ale la feosa—costwt+e)} taking the commencement [cosa—cos(a+y)] of the fall of current as zero axis 1 7 2 revi? . dots [78 : dot +1(——7)] r.m.s. current. ao 0 3 The solution of this expression is: T=1y. V3U—3 oa, Y)).- eee eee (3.23) h yd Sinl2-+e0s(2a-+7)]~y[1+2.¢08 a-+005 (a+7)]! where Y(a, y =a, [cos a—cos (a+-y)]? | sain lichen, (tee eee 46 BRIDGE RECTIFIER AND INVERTOR PARAMETERS lp=ld [/- {cose ~cos(e<+ wt] [cose<-cos(o<+7)] b= Td [cose< ~cos(wt+e<)] [cose< -cos(<+7)] ( ég =dd (4) 1-00 was ie os RKEE EE 0-97 44 Sy IN | cc =O g 0:% a ~— rN o< 2/04 | m 0-95 eto L— ; 3074 S 0:94)25,760°~ 90° 0-93 7 0-92_| 0 10 20 30 40 50 60 T ° (0) Figure 3.19 R.m.s. valve of the transformer secondary current, taking both commutation ard delay angle into account. (a) Division of the current waveshape into 3 parts. (b) True r.m.s. current@s) [For invertors substitute § or (8—y) for al PO ROI ee SS R.M.S. TRANSFORMER SECONDARY CURRENT 47 Figure 3.19 (6) shows the value of 1/[1—3. ¥ (a, y)] for different values of a and y. If it is assumed that the current during commutation rises or falls at a constant rate, i.e., assuming a trapezoidal shape for the current wave, the r.m.s. current will be: ler Ji _% 4° 30 32, yin radians ..........00 000 e eee (3.25) Hence the power factor, obtained by equating d.c. and a.c. power: Vy. V3.E.1 Vols [cos a+ cos (a-+y)}. I, 7 2 V3. 5757 Vee jets [03.4@n1 3 [cos a+cos (a+ y)] Pe UB | Meee (3.26) Power factor is discussed further in Chapter 10. A=cos ¢= 3.10 Converter equations in terms of the per-unit a.c. reactances Usually the per unit reactance drop at full-rated load for the complete equipment is known. It is thus common practice to consider the inductive effects in the converter equations in terms of per unit reactances, instead of the inductances L as has been done in Section 3.2 ~ If r,) is the rms. full-load, a.c. rating of the transformer secondary then the per — unit inductive reactance y teh _V3 0k Ney pu Exnase E . 2 — . . : From equation (3.2), Tet) -/% Tarr) Where Tyzj) is the full-load d.c. rating corresponding [ | to [r,) (assuming a rectangular current wave); from equation (3.1), B=55 Vy | y,, Sel lure { 7 Vy . . . . 3wL . Equivalent commutation resistance —— [equation (3.7)] - 7 " 2 "Tue . Buk - V, 8 L, 2 sV= 7 = Xow ° a or V Xu a 3.27 om 7 2 Tart) Vi, 2 Tygpy ttn tne eee eee ( ) E f i ' 48 BRIDGE RECTIFIER AND INVERTOR PARAMETERS . - Ig. V20oL . . [ In Figure 3.7, the horizontal axis -% Vv. when expressed in terms of the per unit react- . I, { ance will be —¢ , Xue Tari £ The per unit reactance Xu Will consist mainly of the transformer leakage reactance Xropuy and the system reactance X, ‘sie If the short-circuit capacity Ps of the system at the con- +o: _P A . verter terminals is known, then X seo where P is the converter rating. s ' : i 3 act- = t co) Xe [ co os i ada . CHAPTER 4 Basic Requirements of Grid Control References : (3) (10), (39) to (42) inclusive. 4.1 General It is well known that in mercury arc valves and thyratrons, for each value of anode voltage with respect to cathode, there is a value of the grid voltage in relation to the cathode at which the valve will start to conduct. When a positive voltage is applied to the anode, the valve will not conduct until the grid is raised to or above this critical grid voltage. At this point the valve strikes and is said to fire. Once the valve fires the grid loses its control even if the grid is made negative again. The grid has no more control since it is surrounded by positive ions. The current can then only be interrupted by bringing the anode voltage to zero or negative. In a.c. circuits this happens automatically after half a cycle. Thus by giving the grid a negative voltage, the valve can be prevented from firing, and it can be made to fire by making the grid positive when required. Complete control of output voltage from the maximum value to zero can be obtained. Various ‘methods of control have been designed. While designing the grid control system for mercury arc valves, it has to be borne in mind that this critical grid voltage, though supposed to be constant, is by no means constant, since it is subject to tempera- ture and vapour density. Though by a proper design of valve the temperature and the vapour density are kept constant, the temperature and vapour density of a particular Spot is very variable, due to erratic movement of the cathode spot (are path) which also results in jets of mercury vapour (Chapter 12). Thus the only method which is really suitable for accurate control of mercury arc valves is the “pulse” method. 4.2 Pulse method of firing In this the grid is kept permanently negative and a pulse is applied at the required instant. By shifting the phase of the pulse, the firing can be obtained at any required angle and by having a sharp steep pulse, very accurate control can be obtained irrespective of fluctuations in the critical grid voltage characteristic. There are three types of pulses: (a) Peaks. (Impulses of very short duration.) [Figure 4.1 (a).] (6) Pulses lasting for some time. [Figure 4.1 (6).] (c) Pulses lasting for all the time for which the valve conducts. [Figure 4.1 (c).J ‘For bridge circuit these pulses will be of 120° duration, since each valve conducts _ for 120°. Long pulses have certain advantages over short pulses. (i) Suppose short pulses (peaks) are applied to the bridge circuit valves as shown in Figure 4.2 (a). To start a bridge circuit two valves must start simultaneously. This does not happen in this case. To start, an additional pulse can be given later to the valves flag, ot oo Sonos 50 BASIC REQUIREMENTS OF GRID CONTROL Y% Anode voltage 0 Vy! A _f] Grid pulse (a) (6) (c) Figure 4.1 Types of grid pulses after about 60° as shown by one dotted pulse in Figure 4.2 (a), but this means further complications. Starting can also be brought about by making the pulse greater in duration than 60°, as shown in Figure 4.2 (6). To start a rectifier and invertor system, two valves of the rectifier bridge and two of the invertor bridge must fire simultaneously; and if the rectifier and invertor are not syn- chronised to give a simultaneous start to all the four valves, both of these methods will fail. Though methods of this type have been devised by which the system can be started, they involve circuit complications as well as delay in starting. These methods also introduce 150 c/s harmonics during starting; if the line is short [that is, if the natural frequency of the line capacitance and inductance (mainly consisting of smoothing inductance), approaches 150 c/s] it is dangerous to adopt these starting methods and makes the line liable to over-voltages.@ 4 By making pulses of 120° duration, that is for all the conducting period of the valve, both the invertor and rectifier can be started at any instant without any need for synchronisa- tion as can be seen from Figure 4.2 (ii) The pulses of 120° duration (for all the conducting period) are continuous and even it by chance the arc quenching takes place in any valve, it will restrike due to a continued grid pulse. Such reasons as those outlined have resulted in 120° pulses being adopted for the control of valves in h.v.d.c. systems. 4.3 Phase shifting The control of the phase of the pulse is commonly obtained through the use of phase shift circuits: these provide an alternating voltage output, constant in magnitude and con- trolled in phase with respect to the input alternating voltage. The output from the phase shifting circuits is applied to the pulse generating units. Some of the methods for providing phase-shifted pulses are considered below. (a) Figure 4.3 (@). A fixed capacitance C, and a variable resistance Ry are connected across a resistance Ry. Input is given across R, and output is obtained from the mid | PHASE SHIFTING ; 51 / 3 5 é y 8 RY8B denotes transformer phases /,2--6 denotes valves 6 2 4 iM ‘ / ie AS ) ‘jh? \4 \é / 3 sf] () 2 4 6 : = = Rectifier (c) 6 2 4 6 Z a Lnvertop / 3 5 Figure 4.2 Starting of bridge rectificr and invertor in relation to different types of grid pulses point of R, and junction of C, and Ry. It can be seen that by changing the resistance Re the phase angle of output voltage Veg is changed with respect to Vy, and the magnitude of Veo is always fixed at V4,/2, the radius of the semi-circle. The variable resistance will have to be an automatic servo arrangement in actual systems, to give the required change in phase angle according to the programme of regulation and Protection (Chapters 5 and 6). Re, can be simulated by an electronic valve. (6) Figure 4.3 (6). In this method a fixed resistance Rez and a variable inductance L are connected across the resistance R,. In similar manner, the phase angle of Vgg can be changed with respect to V,, by changing L, and it will always be equal in magnitude to Vgy/2. The induetance will be of the transductor type, the value of which can be varied by varying the d.c. through the transductor. This method is better, since a transductor is mechanically and electrically more reliable than valves, and has a reliable characteristic. To obtain the required phase shift, the d.c. in the control winding of the transductor will be Varied by the regulator in accordance with system requirements. It is necessary to feed the outputs to buffer amplifiers, of high output impedance. A a em las ali nmi eA OS te pn ni 52 BASIC REQUIREMENTS OF GRID CONTROL Q i R2 ve — C i Output t Ri i Figure 4.3 Method: A 8 AP B of phase shifting» 2 AB ib Input (a) | c f Control i ; lc, Ly, Output C. t ' A A oa 8 |e: ne { Ce) The above two methods are useful for phase shifting from 0° to 180°, vide continuous control of the invertor from rectification to inversion. There are many variants of this type of phase shift control. (9 42 Figure 4.12 shows one such arrangement, the one used in the Trollhattan-Mellerud experimental plant in Sweden.“ The regulator for constant current regulation is also shown (Chapter 5). In order to produce satisfactory control of bridge circuits it is necessary for the phase- shifted outputs of these circuits to be converted into pulses of 120° duration. (This is discussed in Section 4.4.) (c) Computer method. The above methods are quite accurate enough for the regulation and control of the rectifier, but are not accurate and fast enough for invertors, in which accurate control of firing angle is very important from the point of view of safety of the invertor. Though safety can be ensured sufficiently by maintaining a large angle to give a safe margin, even during faults on the a.c. side, it involves a waste of reactive power (Chapter 6) and poor utilisation of the equipment. To maintain an invertor in safe operation, it is desirable to compute the exact safe firing angle of each valve Separately, so as to ensure correct firing irrespective of abrupt changes . in voltage and current. This will also ensure that the consumption of reactive power will be a minimum. There are three basic requirements: (1) Continuous calculation of the available angle, or available area 84 for commutation. (2) Continuous calculation of the required angle for safe commutation, area. and can thus pro- or the required (3) Comparison of 1 and 2 above, and the provision of an output in the form of an impulse when the two angles are equal. ARRANGEMENT FOR PULSES OF 120° DURATION 53 Thus, using a method of continuous computation, the exact angle of firing is calculated separately for each valve and an output impulse provided when that angle of firing has ( arrived. These impulses are then converted into pulses of 120° duration and applied to i the grids of the relevant valves through pulse transformers. 4.4 Arrangement for converting a sinusoidal yoltage into pulses of 120° duration ods The following are the requirements for a pulse producing unit: 8 (i) The rise and fall times of the pulse should both be very short, this ensures that the pulse shape is rectangular and that there is no delay in starting or stopping the pulse. i (ii) The pulse for any one valve should be switched off coincidentally with the starting i of the pulse for the next valve in sequence. This feature is very important for invertors, since its non-fulfilment could result in commutation failure. (iii) There should be provision for blocking, which should preferably be arranged to stop all valves simultaneously. Figure 4.4, which is a laboratory arrangement used by the authors, illustrates a system for controlling one valve. 7, and 7. are two halves of a cathode coupled binary circuit; in this well-known circuit, T, conducts when Ts is cut off, and vice versa. When 7; 2 is conducting, the voltage drop across R3 maintains the common cathodes at a positive potential. The potential at gy is slightly more positive than this because of the non- conduction of 7, and the choice of values for Ry and R;; conduction through 75 is thus sustained. Bias for T, is obtained from the potentiometer Re;_the voltage input from the phase-shifting device is applied in series with this bias. Referring to Figure 4.5, when the grid voltage of 7, crosses the cut-off line at point A, T, will start to conduct and the potentials of points P and &: fall, thus reducing the current through 7, and tending to lower the potential of the common cathodes, which in turn makes 7, fully conducting. 4 hi a 400V R Pulse Qh /ransformer 1 — R7 | bias 73 Figure 4.4 Electronic unit for producing grid pulses 54 BASIC REQUIREMENTS OF GRID CONTROL Pulse Cathode voll Cut off V. Grid- bras V. A 8B C OV. ; OV. Figure 4.5 Operation of a pulse unit The effect is cumulative and the transmission from Tz conducting to T, conducting is very rapid; T, remains cut off and T, conducting until point B (Figure 4.5) is reached somewhat later in the same half-cycle, when the reverse process takes place and initial conditions are restored. The relative durations of the two intervals AB and BC may be controlled by variation of the bias applied to gy. TX possible output arrangement is also shown in Figure 4.4, the voltage across Ry being applied to the cathode-grid circuit of T3. When T. 2 is conducting, the voltage across Ry serves to keep 7; cut off; output in the form of a pulse of duration AB is thus available when T, is conducting. An arrangement such as this does not accurately satisfy the requirement (ii) above. Reference to a more complete diagram of the control circuit of a model bridge circuit, as in Figure 4.6, illustrates the additional circuit interconnections which are necessary. The grid of T3 of one unit is connected to the anode of T, of the next unit through a resistor, e.g., Ri, and a metal rectifier; then considering valve 3, its pulse will start when T, of its control unit starts to conduct. But conduction of T, causes a lowering of potential of point P which in addition to cutting off Tz of its own unit, now applies negative voltage to the grid of 7; of the unit controlling valve 1. Similarly, the starting of the pulse for valve 5, cuts off the pulse for valve 3, etc.; the control units for valves 2, 4 and 6 form an identical closed circuit arrangement. With such a control system, only one bias potentiometer Rg is provided as shown in Figure 4.6. The blocking of all units is accomplished by short-circuiting the bias voltage, ie., closing the switch Y-Y, and bringing all valves T, to cut-off. The necessary condition for such blocking is that the peak value of the applied a.c. nets os liso aii ARRANGEMENT FOR PULSES OF 120° DURATION 55 4] connecl2d sef | | c regulator Phase shifting [ { FE + 2 eee Tete! I, ‘Ses [nas gD ara T aD om, foo Figure 4.6 _ Circuit volves 2486 arrangement for con- verting phase-shifted voltages to pulses of 120° duration for use in bridge connection © Control unit | xi ly - Blocking terminols voltage is somewhat less than the cathode voltage minus the cut-off voltage of 7); this is evident from Figure 4.5. An alternative way of blocking may be by short-circuiting the anodes of valves T2 through a resistor Ry as shown in Figure 4.4. This causes current through Rz and Rg in series, and the voltage drop across Rz cuts off Ts and blocks the pulse output. 4.5 Arrangement for converting short impulses into pulses of 120° duration ‘ The requirements in this case are as in Scction 4.4 (i), (ii) and (iii) above with the addition | that if the control pulse is to be started by positive impulses, then it should be unaffected by negative impulses and vice versa. Figure 4.7 shows an arrangement used in the control of the Gotland h.v.d.c. scheme.@0 The roles of the valves 7), Tz and T3 are the same as those discussed above in Section 4.4; Ry, the anode load of To, is smaller in value than R3, and thus the currents through 7; and 7, during their respective conducting periods are dissimilar. When T> is conducting the cathode voltage is 15 volts and the grid of T, is clamped at 7 volts by diode 1 and so remains cut off. 7 If a positive impulse is now applied to the grid of 7, through the resistor Rj, its potential will rise above 7 volts and the valve will start conducting; in the usual way the fall in voltage on the anode of T, cuts off 7, by means of the potentiometer R;-Ro. Similarly, as T, cuts off, the voltage on its anode rises and, through the potentiometer R;-Rg, the grid of Ts attains 50 volts and starts to conduct thereby starting the pulse. Under these conditions, the voltage across R. becomes 7 volts, which is also the value to which the grid of T; is clamped; thus any negative impulse reaching the input will have no effect on ¥; and the pulse continues once it has been started by a positive impulse. When T; is not conducting, the voltage across Ry is zero; when it is conducting, this same voltage is clamped to 50 volts by diode 3. This is utilised in the arrangements for ensuring that-the stopping of one pulse is coincident with the starting of the next. The voltage across the Rg resistor of the next rectifier to fire, rectifier 3 in this case, is applied to the R, resistor of the preceding unit through diode 4. Thus as soon as the T3 valve i { wie ae & a ° iat rT 2 = cuy tical [ 4 4 nts e ea) gl é on i tina = SS 56 . BASIC REQUIREMENTS OF GRID CONTROL i ube \ Pulses to Pulses to | y— Pulses bo ' transformer IL valve f Me vate AN vole 5 +450V I Ty i ? Bes fey ' a a ms Units for [+ 3B = % r2 valves 2466 % [_| ei +0 t 5 3 +50 C +50¥ Is | 2 3 i ‘ R R i ow 0} [ 9 ie 9 ov i + ; i Rs Rs } -600V 6 8 ~600v 6 é Figure 4.7 Circuit arrangement for converting impulses to pulses of 120° duration for use in bridge-connection of rectifier 3 starts to conduct, 50 volts is applicd to the Rz resistor of the unit controlling rectifier 1 which cuts off its T), starts its T, and cuts off its T3. Additional control is exerted by diode 2; this limits the upper voltage attained by the grid of T, to 22 volts, thereby ensuring that a positive impulse applied through R, cannot siart rectifier 1 when the pulse controlling the following rectifier, i.e., rectifier 3, is switched on. When the pulse of the following unit is switched off the potential across R, becomes 15 volts because of T. conducting, and the grid of T; is again ready to receive a positive impulse when commutation takes place from rectifier 5. Thus the three odd-number rectifiers in a bridge circuit are controlled by the following sequence of events: (i) Pulse for rectifier 1 starts and cuts off the pulse for rectifier 5; control circuit of rectifier 3 is quiescent and ready to receive an input positive impulse to the grid of its T, valve. (ii) Application of the input impulse to the unit controlling rectifier 3, starts its pulse and cuts off that from the control unit of rectifier 1; control unit of rectifier 5 is now ready, and so on. The control units for rectifiers 2, 4 and 6 form an identical closed circuit. The cathodes of all the 7, and T; valves are connected to a busbar b through diodes 5 as shown. When the voltage of b is raised to 50 volts, all the valves T; are cut off thereby blocking all control pulses. The diode 5 serves to isolate all the control units during normal operation. 4.6 Grid bias and other auxiliary supplies To maintain the rectifier valves in the bridge circuits blocked, direct voltage has to be applied to each grid, negative with respect to cathode. Three of the six rectifiers or in- vertors of a bridge connected set have the same cathode potentials and it is possible to use a single bias unit for these. It is preferable, however, to have separate bias arrangements for each rectifier of the set since this will avoid capacitive interaction between the grid circuits of those having the same cathode potential. : The most suitable way to apply d.c. grid bias is to supply a.c. through transformers, [rap en [ GRID BIAS ARRANGEMENTS AND AUXILIARY SUPPLIES 57 <i | F + Ebb r Pulse | fransformer | Bridge ei \ t clreult E a | Erie g, (4) : Exponential decay ae __ (hime constant ‘mp ) Ep YT | Fi | | : Exponentig! decay (time constant “m/e, approx) (6) Figure 4.8 Basic principle of applying pulses over the grid bias, and shape of the grid pulses tm a4 ; Wit skh insulated to the appropriate cathode voltage, and then to rectify the a.c. by means of dry- type rectifiers mounted on the valve chassis. Valve auxiliaries (ignition, excitation, etc.; Chapter 13) may be supplied in a similar way. A further d.c. auxiliary voltage, but at earth potential, is also required for the operation of regulators, pulse units and other control equipment. It is essential from the point of view of safety that this auxiliary supply should be reliable, and that it should be maintained in the event of a failure of the main a.c. system E (Chapter 6). Thus, even though auxiliary supplies are obtained from the main a.c. system c during normal working, it is very desirable that an alternative source should be available. 3g G Sec /. : (Eine 3 a 58 BASIC REQUIREMENTS OF GRID CONTROL A convenient arrangement for this purpose is to instal a diesel-electric generator or a set of batteries for the d.c. supply, together with a small motor-generator sct, fed from a battery source, for the a.c. supplies. 4.7 Pulse transformers These serve to isolate the control-pulse producing equipment from the high voltages associated with the valves in the bridge-connected sets. For the reproduction of pulses without appreciable distortion, it is important that the leakage inductance of the pulse transformers should be very low and the magnetising inductance high. This latter feature is most important for the type of control pulse at present being discussed since these re- require a long time constant, and may be obtained by using a large number of turns and high permeability material for the transformer cores. Compromise is necessary, however, since the large number of turns tend to increase the leakage inductance to unfavourable Proportions, as well as increasing the shunt capaci- tance of the winding and so causing deterioration of the steep sides of the pulse. Detailed design considerations of pulse transformers may be found in a number of published books on pulse techniques. 4.8 Arrangements for impressing pulses over the grid bias of individual valves in a bridge- connected set The basic idea for accomplishing this is shown in Figure 4.8 (a). The valve T3 of this illustration corresponds to the valves T; of Figures 4.4, 4.6 and 4.7 and controls the stopping and starting of the control pulse; the shape of this, as obtained from the secondary side of the pulse transformer, is shown in Figure 4.8 (5). E, is the initial height of the control pulse and depends mainly on the turns ratio of the pulse transformer, the transformer load resistance Ri, the voltage £,, and the plate characteristic of T3. The exponential decay of the pulse is governed by a time constant of approximately L,,/R, where L,, is the magnetising inductance of the pulse transformer and R is the parallel combination of Ry, Ry and the plate resistance of the valve T3, all referred to the primary side of the transformer. Ry is a grid current limiting resistance and hence is large; thus +£54 Ro ‘ a sven Suite Pulse a 2 ¥ / circuté transformer Ze 3 34 Sl way 73 \g a Figure 4.9 Method of applying pulses over the grid bias cn anne i tana ARRANGEMENTS FOR IMPRESSING PULSES 59 1454 ; Fo Bridge Pulse g R3 circuit /ransformer S\|8 5 Ry f {| ar / V, mega i 9g Figure 4.10 Method of applying pulses over the grid bias R, should be made very small and the valve should have a small plate resistance in order to obtain as flat a top to the pulse as possible. After a time t, the pulse will have reduced to a height of E,’ according to the value of its time constant. Under normal working conditions in bridge connection, the time ¢ should correspond to 120°, which is equivalent to 6.7 milliseconds on a 50 c/s basis; during a sudden phase shift, however, one of the bridge circuit rectifiers may have to conduct for a longer time than this and a design of pulse transformer and associated circuit should be considered which allows for a time corresponding to, say, 200° or 11.1 milliseconds. When T; is cut off to terminate the pulse, the dissipation of stored energy in the trans- former inductance causes a pulse of negative magnitude E,"’ to appear. The decay time constant will be approximately L,,/R,, and decay will have to be complete within 240° for normal operation, or as little as 160° during a sudden phase advance; this raises a thy Ro Bridge circuit Pulse | transformer +4 k- - SL www, 73 Figure 4.11 Method of applying pulses over the grid bias 60 ‘ a 4 a = | 3 q : e \ ; s “ + hii 2 | afle————_| 5 ae Oe eT ARRANGEMENTS FOR IMPRESSING PULSES 61 difficulty since R, is normally made small to obtain a long time constant for the positive control pulse and will consequently give a longer than desirable decay time for the un- wanted negative pulse. The simple circuit of Figure 4.8 (a) although illustrative of the " basic arrangement, is thus impracticable for use with bridge circuits. Metal rectifiers may be used as shown in Figure 4.9 to provide separate conducting paths, and thus discriminatory time constants, for the positive and negative pulses. Thus, during the operative (positive) pulse, conduction is through diode 1 and R; may be made sufliciently small to give along time constant. Diode 2 conducts through R3 when the pulse is negative, and Rg is made sufficiently large to ensure quick decay. One further difficulty is that the larger the value of Rg, the greater will be the magnitude of the negative voltage peak. Thus at the time of switching off the control pulse, a large voltage, £,’’+V,, will appear across the grid and may well be undesirable; Figures 4.10 and 4.11 show suitable improvements. . In the arrangement of Figure 4.10, the negative voltage drop across resistor R3 appears across diode 1 and not between grid and cathode as in the previous case. Diode 2 may be omitted from this arrangement so that current flows through both R,; and R; during the positive pulse but only through Rg during the period of the negative pulse. In the circuit of Figure 4.11, the current flows through R, and diode 1 during the positive period, and through Rg and diode 2, against the bias V, and diode 3, during the negative period.4” In the pulse transformer, the magnetic flux rises from zero to some positive value during the positive part of the pulse and reduces to zero during the negative pulse. Thus the core is only magnetised in the positive direction and is not fully utilised; if, however, d.c. is passed through a third winding on the core, which magnetises the core in a negative sense, the core can be fully utilised and its size greatly reduced.¢ Figure 4.12 Phase-shift device for the grid control (a) Schematic diagram. (6) Complete diagram including the devices for automatic current regulation. S, three-phase supply; 7, trans- former; X, transductor with ele- ments X, and X.; R, resistance; Eg, output grid voltage. The main current to be regulated is transformed by the measurement 7 transductor Xm, rectified by Lm and loaded on resistance R. The voltage over this resistance is com- pared with that of the potentiometer Pand fed into the grid of the triode E. This triode controls the excitation Current of the phase-shift trans- ductor X. The output voltage E controls the ignition point of the main converter valves via a set of electronic valves and insulating transformers. F CHAPTER 5 Compounding and Regulation References: (3) (12) (41) (43) (44) (45) (46) (47) (48) (64) (106) (107) (108) 5.1 General) The term “compounding a converter” implies selection of converter characteristics in order to meet the requirements of regulation and protection. From the characteristics indicated earlier, it is seen that Vand I, vary in terms of a and y for the rectifier, and B, y and § for the invertor. By varying these quantities in a suitable manner, the rectifier and invertor may be given any required characteristics over the whole range of operations. Suppose it is required to arrange the rectifier regulation in such a way that the d.c. voltage at the end of the transmission line (invertor end) remains constant. The rectifier voltage at the end of the transmission line is given by: V,=V, cos oR) WT Bal, . 3 A 7 : : 7 where is the equivalent resistance for commutation drop and R is the line resistance. T Let Vj;=V,=constant, then: 3e0L, Vi cos a—Iy, —29=V.41,R=a+bIy. oo. ccc ecc ccc ee (5.1) wT The left-hand side of equation (5.1) is the d.c. output voltage at the rectifier end and can \ be measured. Thus a rectifier regulator is required which takes into account the d.c. ‘output voltage and current, compares the measured output voltage with the computed / , Value from the right-hand side of equation (5.1) and provides an output for any difference between the two. This output will be fed into a phase-shifting device which can change ithe phase angle in accordance with the difference output until the steady-state is reached and the above equation is satisfied. ~ The invertor can be similarly arranged. Thus, in theory, the whole system can be organised to give constant current, constant voltage or constant power regulation as required, 5.2 Required regulation It is important that the valves should be operated strictly within their current ratings, since there is a substantial risk of damage if the current is increased beyond the rated value for even a short time; constant current regulation is thus clearly desirable. If the reduction of invertor back voltage or the occurrence of a fault on the d.c. line tends to increase the current to a large value, the rectifier grid control should rapidly increase the angle of fire, and thereby reduce the rectifier output voltage to such a value that current is maintained constant, or kept within prescribed limits. aoe staan pane ie nas a in endian ent ne aN a at ins it noi ninlnsitrt, INVERTOR COMPOUNDING 63 Furthermore, a change in the power transmitted in accordance with the requirements of the receiving system is also necessary. If the receiving system is large compared with the d.c. link, the d.c. link may only have to feed constant power; if the d.c. link is the only source of power, it may have to carry out the power regulation in accordance with load requirements. F 5.3 Invertor compounding 7 The time, or angle of advance, required by the invertor is given by the equation: l= z (cos §—cos f) (3.18) A TD wleg 008 8-008 B) «0s ee vere eevee esee ness : hle ata B=8+y LE Each valve must start to fire at an angle f in advance of the voltage zero so that there is iad : still an angle 8 remaining after the completion of commutation. Angle 5 has to be at least equal to 5, the deionisation time of the valve, and may be approximately 10°. If this [ condition is not fulfilled, the valve which has just ceased conducting will conduct again and produce a commutation failure (see Section 6.4.1).~ : Thus the operation of the invertor is associated with the consumption of a certain mini- f mum amount of reactive power, and the greater the angle 8, the greater will be the reactive tag power. For known values of E, /,, 8,, w and Ly, it is possible to calculate 8. The current and voltage are liable to considerable changes and hence some extra angle has to be pro- rm vided to take account of these changes; in this way reasonable safety can be provided. f) : “ Safety obtained in this way is clearly at the expense of reactive power; the greater the factor of safety, the greater will be the angle and the reactive power consumption. Even dycan so there is a substantial risk of failure for the following reasons: [ie \ (i) A sudden fall in the output voltage of the invertor can occur as a result of faults or pitted . sudden load changes in the a.c. system; this tends to reduce the commutating rence voltage and hence increase the required commutation angle, as shown by equation lez (3.18). csd In addition, a phase to phase fault will not only reduce the commutating voltage i but will also decrease or increase the available commutation angle. This can be seen from Figure 5.1 in which the reduction in voltage of phases R and B has resulted in an increased angle of commutation for commutation from valve 6 to valve 2, and from 3 to 5, and decreased the available angle for commutation from valve 4 to valve 6 and from 1 to 3. 7 (ii) When the invertor a.c. voltage drops, the invertor d.c. voltage also drops and the , 4c. line capacitance discharges through the invertor; in this case the invertor curreat will rise momentarily even if the rectifier has been given a constant current characteristic. These considerations demand a considerable angle of advance to mect reasonable safety requirements, and yet every attempt must be made to economise in the amount of reactive power considering the large amounts of power involved in h.v.d.c. projects. In order to compromise, the invertor should be compounded in such a way that suitable variations in & 7 co 7. “” Q 3 Gq nag G = —] sam ww TES / magnitudes may be in the form of d.c. voltages, both 64 Increased Reduced angle angle Figure 5.1 Changes in commutation voltage and angle for commutation, due to changes in a.c. voltage the angle of firing are provided in accordance with instantancous values of voltage and current. This involves the use of a computer which can calculate exactly, at every instant, the angle which is required. Such an arrangement has been used in the Gotland scheme and given the name “consecutive grid control”.4) The complete arrangement consists of a Separate computer for each valve, which con- tinuously calculates and provides an output pulse when the correct minimum angle of firing has arrived for safe commutation. This output pulse starts the pulse of 120° duration (as described in Chapter 4) which in turn controls the appropriate valve. To ascertain the exact angle of firing, two magnitudes must be compared; one of these ‘is directly proportional to the required angle and the other to the available angle. The : applied to a comparator which gives an output when they are equal. This process involves continuous knowledge of the commutating voltage, its phase dis- placement and the direct current magnitude. The commutating voltage is the voltage prevailing between any two valves which normally fire in succession provided that, at the instant in question, one of them is non-conducting. Before commutation this voltage appears as a voltage across the valve which is to strike; this is also the v oltage between the phases to which the valves are connected. Thus the commutating voltage for commuta= tion from valve 3 to valve | is the voltage between phases Y and R. It may be possible to carry out compounding in the following way: i 4 INVERTOR COMPOUNDING 65 =-R, ald 4“ 7 Vy=Vocosd,-3Wla ly [ +f [ Figure 5.2 Characteristics of a compounded invertor By re-arranging equation (3.18), and putting 5=5,, V2E cos B= +/2E cos 8,—2wLely oo. c eee c cece cece eee eee (5.2) The right-hand side of this equation may be compared with its left-hand side, an impulse being provided when they are equal; it should be appreciated, however, that comparison is made with instantaneous values only. ! The instantaneous measured commutation voltage is 1/2E sin (180—B)=1/2E sin p. [ It is clear that the left-hand side of equation (18.1) \/2E cos f can be obtained from either : the rate of change or integration of +/2E sin B. ana On the right-hand side, cos 8, is constant where 8, may be taken as 10° to 15° as required. ant, V2E may be taken as the measured peak voltage at every cycle; otherwise, it must be . { computed. J, is given by the instantaneous measured current. Le cot Thus when the continuously calculated values on the two sides of equation (18.1) are equal, the resulting output impulse from the comparator will initiate the control pulse for the corresponding valve; such a method of compounding may be termed “constant_ex-_ tinction angle (8,) compounding”. | Account should also be taken of the rate of change of direct current, since the change of this quantity during the actual process of commutation also changes the commutation angle. A compounding device working to these principles may not be exact under all circum- stances, but can be made exact under normal working conditions when the commutation voltage is following a sinusoidal variation; with some empirical adjustments it can be made to respond correctly to transient phenomena arising as a consequence of disturbances in the a.c. system. With such.an arrangement, commutation failure is only likely when changes occur during - the interval when commutation is taking place. Even if commutation failure occurs, the invertor will automatically resume correct operation if if the next valve operates correctly (see Section 6.4.1). As the direct current increases, the angle f will increase and hence the invertor d.c. voltage will decrease. Thus the invertor characteristics will be as shown in Figure 5.2. mt Ge z fe SC 2 sa a oe eS ne Lie it ‘ sabe sete say! ES te oe wa a Rael Ca na a ice Los COMPOUNDING AND REGULATION current, the invertor back voltage will be: (eZ 35 Pa Ecos &, when the angle of lead is only 8,, and will change with a negative slope, i.e., it will reduce with a slope of: ~ 3 aL. REL Bed etalsla geese add teil (5.3) TT In practice, the slope may be somewhat greater. 5.4 Uncompounded invertors When the invertor is not compounded, there is no provision for changing f and this angle will have to be large to avoid commutation failure during conditions of large direct current or reduction in invertor voltage on the a.c. side. The equivalent circuit and characteristic 2 of an uncompounded invertor will be as shown in Figure 5.3; is maintained constant and, at zero current, the d.c. voltage is (3\/2/m) Ecos 8. This voltage incre current, at a slope of 3wLo/a= Re., o (4 |__—————__— Pr ertee 3wlo Rk Le PU At eat tay ___—_~ SN MTT ee Ty VY, cosB+3wly ly dy Figure 5.3 Characteristics of an uncompounded invertor 5.5 Rectifier compounding § For the reasons indicated in Section 5.2 above. t a constant current characteristic, The perfect constant current characteristic will be a vertical line as shown in Figure 5.4. Thus for by shifting its firing angle, constant. ~ , itis desirable that the rectifier should have ! at that current setting, any invertor back voltage, the rectifier will adjust itself to give a d.c. output voltage such that the current will remain This can be accomplished by a current regulator which compares the actual current represent both currents by means Wo, the angle a will be changed with a set value of current; in practice, it is convenient to of d.c. voliages. If there is any difference between the t in the correet sense until the difference is zero. zero ‘educe [’ (3.3 agele rf risac tand, 6 n { 5 [Tree Seremgneen mOae ee menryeramtrmternemuremnein icf = RECTIFIER COMPOUNDING 67 Vy a Y { Vays Vocos cc-(3wheR)lyy 0. 7 ey tie “30E el / Vocosoc- (Bul Ry Vy cos of = Myo Vcasce Sully T fat tp las——ly Figure 5.4 Ideal constant current rectifier compounding The top of the characteristic is the rectifier output voltage when the delay angle a is zero; it may be called the “natural-voltage characteristic”. The voltage from this natural- voltage characteristic is given by Vj=34/2E/7—(3wL,/7) . I, where (3wL,/7) . 1, is the drop in voltage due to commutation. The voltage V, will increase to V,=34/2E/zm for J,=0; thus the slope of this faa voltage characteristic is: Vi—Vqg_ 30, _ miei The characteristic may be shifted vertically by changing the value of £ by selecting transformer taps; vertical shifts of the characteristic will naturally occur with the fluctua- tion of E. The slope of the ideal constant current characteristic is given by: Pe Be by stelelerclelelt dake tb bel plnlorele ald acs (5.5) 81, Such an ideal characteristic would be very unstable. It has been shown by Busemann | that owing to the negative slope of the invertor, the rectifier characteristic must have a minimum slope in order to avoid hunting during sudden d.c. voltage changes. The cause of this hunting is the oscillatory circuit formed by the line capacitance and the inductances of the stations and line. The equivalent circuit representing the d.c. system is shown by Figure 5.5 (a). V,, and Vs are the no-load direct voltages of the rectifier and invertor respectively; C, is the d.c. line capacitance lumped in the centre; R’, is the total resistance of the rectifier side consisting of the slope R, of the rectifier constant current characteristic and half the line resistance R ;(—R’.) is the total resistance of the invertor side, consisting of the slope (— R2) of the compounded invertor characteristic and half the line- resistance R (R's will be negative when R2>R/2); L'n and L’z2 are the equivalent reactances of the rectifier and invertor side respectively, each consisting of the inductance of the smoothing reactor, half the line reactance and the effective inductance of the a.c. side which may be taken approximately as the inductance of two phase windings in series. (44) = > f ss om fa Gh Gmr mz 68 COMPOUNDING AND REGULATION R> Ly Lin ~Ro VW OH004—O00 — po Figure 5.5 Hunting conditions for the d.c. system In such a circuit the d.c. at the rectifier is not necessarily identical with the d.c. at the invertor, since an oscillatory current may be superimposed which will flow in the circuit during any sudden changes in the voltages. The resist ance values determine whether an oscillation is sustained or whether it is damped aperiodically. To study the effect of a sudden reduction in Vig by 5V i the circuit can be represented by Figure 5.5 (6). If R, [equation (5.5)], and hence R’,, is infinity, it can be represented by an open circuit and the equivalent circuit will be as shown by Figure 5.5(c). In sucha circuit, with nega- i re — = oan ~ om | RECTIFIER COMPOUNDING 69 tive damping even a small value of V2 will start oscillations which will increase and disturb the operating condition. The oscillations will have a natural frequency of: hua J (=z) aaa ep eeaseehtae (5.6) Thus the rectifier characteristic will have to be given such a slope that the resultant damping of the circuit (Figure 5.5[b]) is positive. The current in the invertor, when switch S in Figure 5.5 (6) is closed, can be calculated and will consist of three components, the ‘ Figure 5.6 150c/s hunting voltage due to inertialess regulator and steep characteristic of the rectifier compounding permanent d.c. component, the d.c. transient and the a.c. component. The first will eventually be taken care of by the rectifier, while the latter two will die out depending upon the resultant damping of the circuit. When L'1=L' 2=L', the critical value of R’z and R’, in order that the resultant damping is positive, will be given by: Es) Ly ‘= [{— (Emm cece cee seecececress 27 R's AZ and R’, RC (5.7) Thus to have sufficient positive damping L’, should be larger than C,(R’,)? whereupon , R’, should be less than ba : R2Cy Thus in order to improve the control characteristic of the rectifier, the smoothing reactors should be of sufficiently large size. In addition, the damping can be improved by providing special damping circuits for the corresponding frequencies. The damping can also be; improved if the invertor reactor is larger than the rectifier reactor. This is also desirable; for the control of harmonics, which will be larger in the invertor due to its larger operating: angle, and for limiting any sudden line discharge current through the invertor. | | : | I | C od) Cs fn E E 70 COMPOUNDING AND REGULATION Vy 3V2 7 YI T . Ly Ig lon —»-[ Figure 5.7 Rectifier constant current compounding Other oscillatory phenomena may occur due to the fact that the controller of the rectifier is not continuously effective, but only at each ignition, at approximately 60 degree intervals. A very sensitive controller with a steep characteristic has a tendency to hunting as indicated in Figure 5.6 At point B, the angle is reduced by 8a, since the current fell below the set value and a positive voltage increment was introduced. Due to this voltage increment, the current increases to a value more than the set amount; this influences the controller at the next ignition time C, so that the ignition is just as much too late as it was too early at time B; and a negative voltage occurs which again reduces the current so much that the next ignition at time D takes place too early. It can be seen that the controller thus tends to induce self-excitation at third-harmonic frequency. Investigations of this phenomena have been carried out by Busemann® and he suggests that hunting can be checked by introducing the equivalent of large inertia in a controller with very strong compounding. The more the inertia in the main circuit (large smoothing choke), the less the inertia necessary for the controller. The controller ” can also be given a greater inertia on decreasing current than on increasing current. Thus by giving the rectifier characteristics some slope R,, to avoid hunting, the character- istics obtained are as shown in Figure 5.7 "The current does not remain exactly constant at reduced invertor back voltage but increases somewhat, depending upon the slope R,. Various lines show different current settings, and J,,, is the maximum allowable current setting fixed by the maximum rating of the valves. The slope of the constant current characteristic of the regulator may be increased or decreased by increasing or decreasing the regulator gain. ! | ! | 1 Mt. Id Figure 5.8 Combined rectifier and invertor characteristics with current regulation from rectifier With a current setting of I, (Figure 5.7), the actual current J, fixed by the regulator is changed in proportion to the output voltage with a slope R,. 5.6 Transmission characteristics in the rectifier and invertor compoundiag i ' ' [ : i TRANSMISSION CHARACTERISTICS AS A RESULT OF COMPOUNDING 71 B nase By providing “constant current regulation” on the rectifier (Section 5.5) and “constant extinction angle (8,) compounding” (Section 5.3) on the invertor, the working characteristics of the system will be as shown in Figure 5.8. I In the rectifier characteristics, account has to be taken of the d.c. line drop /,R in order roe oar to match the invertor characteristic. Suppose that the rectifier constant current setting is fixed by the line T,,; when the invertor transformer secondary voltage is set at Z’ the current will correspond to point A. Now if E’ reduces to E”, the current will tend to increase and the rectifier regulator will act to increase the angle a and hence adjust the output voltage te so that the current now corresponds to point B. of A reduction in E may be due to a fault on the invertor side, but the current will increase if 5 \ only slightly. Even if the invertor back voltage collapses entirely, the current will not it ! increase to a value greater than J,,, and will be sustained by a rectifier voltage of reduced te. : value equal to Jy; [((3#L,/7)+ R]. ; : : When the invertor a.c. voltage is reduced, the rectifier a.c. voltage may also be reduced | by means of transformer taps, to bring the upper line of the rectifier characteristic down to | the dotted line corresponding to £’”’, as shown in Figure 5.8. This should be done to avoid operation at a large delay angle a, but care has to be taken to ensure that the upper line of the rectifier characteristic is not lowered below that of the invertor characteristic. If this should occur, the rectifier and invertor will “run down”, the current falling to zero since the rectifier voltage will be insufficient to overcome the invertor back voltage. Provision must obviously be made for normal fluctuations by leaving an adequate margin; at a 72 COMPOUNDING AND REGULATION "this implics that the rectifier should work at a certain minimum angle a. Fast-acting tap- change arrangements should also be provided to increase the rectifier voltage as soon as the angle a falls below the minimum limit. ~ Tt may thus be scen that power transmission can be carried out with adequate safety and reliability by providing “constant extinction angle compounding” on the invertor, “constant _current compounding” on the rectifier, and a quick-acting transformer tap changer. Power _ regulation may be accommodated easily by changing the current setting of the rectifier \regulator. The problem of stability of transmission as affected by the control systems at each end of a d.c. link has been analysed in general form by Reider,“ who drew the following overall conclusions, which are of interest in the light of the discussion above: — (i) The operation of a d.c. transmission line with only a current regulator on the rectifier (no compounding on the invertor) is stable for all values of the angle of lag of the rectifier. ~ (ii) The operation of the transmission line with compounding only on the invertor (no regulator on the rectifier) is unstable and should not be tolerated. 7 (iii) The operation of the system with both control devices is stable. There is, however, a region of unstability which may be reduced by decreasing to some extent the amplification factor of the current regulator. 5.7 Communication link As mentioned previously, the power transmitted by a d.c. link can easily be altered by changing the current setting of the rectifier regulator. It is also necessary to carry out power regulation in accordance with the requirements of the receiving system (discussed further in Section 5.10 below). A communication link is necessary to convey information from the receiving end to the rectifier regulator, and may also be needed for protective (hy Figure 5.9 Current regulation from invertor side + are leila 2-— ScteRometene: EE — We ° e€ ving pan Fam or “woe 2 ama 4 a 1 neo CURRENT REGULATION FROM THE INVERTOR SIDE 73 purposes (Chapter 6), starting purposes, and reversal of the direction of power flow (Section 5.13 below). There are three main possibilities: (i) Short-wave radio link. - (ii) Pilot wires. - (iii) The use of carrier frequencies on the power conductors.” 5.8 Current regulation from the invertor side As discussed in Section 5.6, the current regulator of the rectifier ensures that the current does not rise above the set valuc, but it does not ensure that it will not fall below. This, Rectifier Jnvertor ly Figure 5.10 Combined rectifier and invertor characteristics with current regulation from both sides of course, can be assured by the fast-acting transformer tap-changers which increase the voltage as soon as the angle a approaches zero. It is clearly not desirable to operate tap changers very frequently, and tap changing may not be adequate or fast enough if one of the rectifier bridges is suddenly blocked; this contingency will result in the system “running down” unless an invertor bridge is simul- | tancously blocked. This difficulty may be overcome by providing a “constant current regulator” on the invertor also, which ensures that the current does not fall below a certain given valuc; an arrangement such as this will be quick acting and will have certain other advantages. With such a regulator, the firing angle B of the invertor will have to be advanced beyond) the commutation margin and reduce the back voltage, so tending to sustain the current immediately it falls below the set value. The constant current regulator will be similar to that of the rectifier, but Operating in n opposite sense, i.e., when the current decreases the angle f increases to reduce the back Voltage, / - ' \ oh oes ox chia rie wm ALN Ere 74 COMPOUNDING AND REGULATION For reasons similar to those advanced in the case of the rectifier (Section 5.5), the invertor constant current characteristic will have to be given some minimum slope and this will have the sense shown in Figure 5.9. The various lines are for different current settings, and the upper line represents the working of the invertor at a constant extinction angle. This may be better appreciated by reference to the combined transmission characteristics shown in Figure 5.10, When rectifier and invertor a.c. voltages are E’ and E” respectively, such that the rectifier tap setting is higher than that of the invertor, the rectifier will be running on constant current regulation while the invertor has a safe margin of commutation. This is so because the actual current is more than the current set on the invertor regulator; and hence its regulator output is zero or at a minimum fixed level and so does not affect the phase advance; the current is that given by point A. For working in this manner, the condition which must be satisfied is: 2.£' . E” 2 3V2.E ~1, ( )> 3V2.E cos 8, —/ 3wLo WT wT T as 7 This will be the normal working condition, since operating an invertor on constant current regulation implies a large angle of advance 8, and hence a large reactive power requirement. Suppose (due to a fault or any other reason) that the voltage on the invertor side falls to a value corresponding to E””’; the current will now be given by point B. If, however, the rectifier voltage corresponding to E’ falls to E’”’, then the current will be defined by C; in this case the invertor current regulator has taken over. The Operation now is that with the current falling below its set value, the regulator has provided an output which (in addition to the minimum required commutation angle) has given a further advance angle to the invertor and reduced the output voltage until the current reaches the value corres- ponding to C. Although the current has now been reduced to some extent, this should not result in the system running down; when the current reduces to the set value, a control signal can be given to the invertor-transformer tap changer to lower the a.c. voltage so that the invertor does not run on a constant current characteristic for a long time, thereby avoiding prolonged consumption of large reactive power and possible overloading of the source of KVAr. If tap changing is insufficient then one of the invertor bridge units could eventually be blocked, but this will not be necessary if the reduction is momentary. It can thus be seen that when the invertor characteristic is higher than the rectifier characteristics, the invertor works on constant current regulation and the rectifier works with no delay and natural-voltage regulation. When the rectifier characteristic is higher, the rectifier works on constant current vegula- tion while the invertor runs on the necessary commutation margin. The system does not tun down due to changes in the a.c. voltage or blocking of one of the bridges on any side. Furthermore, it may be seen that in all cases the current setting of the invertor has to be lower than that of the rectifier; the difference in setting is called the “r gin” setting. This margin has to be sufficient to give some difference on the top voltages and avoid both current regulators operating simultaneously; such operation is clearly undesirable and aoe eaten ee TRANSFORMER TAP CHANGING 75 Figure 5.11 -Transformer tap changing may also be very unstable. Thus when the current setting has to be changed, it must be ‘done on both rectifier and invertor regulators simultaneously in order to maintain the margin setting. The high-speed communication link is needed for this purpose; the command for chang- ing the current setting comes from the invertor side according to the requirements of the receiving system. Apart from this ccntrol, the rectifier will have a permanent excess signal equal to the margin setting. Arrangements of this type were suggested by Tcher- vonenkiss“) and have been provided on the Gotland system.“ No augmentation of the reactive power source is necessary since the tap-change arrangements allow the invertor to run on a constant current characteristic for a short time only. The scheme outlined above also enables control of power transmission and its regulation to be carried out entirely from the invertor side in the event of communication channel failure; this is discussed further in Section 5.11. ng ; A further advantage of additional current regulation from the invertor side is that the current through a fault on the d.c. line would be equal to the “‘margin” setting only; both | rectifier and invertor constant current regulators would act to maintain currents on their respective sides equal to their settings. Under such conditions, the invertor current regula- tor would advance the firing angle to more than 90° and thus operate as a rectifier with a voltage output equal to the voltage drop from the invertor to the fault. Limiting the fault current in this way would avoid damage at the fault point and allow for a quick recovery after current interruption. aoe Fe - (in ~™ 5.9 Transformer tap changing The equation of the rectifier voltage at the invertor end is given by: 30L V,=V, cos a—I,\ R+-— T This voltage increases to V,—I, (R+3wL/7) for a=0, and decreases to zero when a approaches 90° and’V, cos a=J, (R+3wL/z). ee oT asa _———— oie Aa that etn nnn ae Ue SS RE nr PRR I sm A: Tras 76 COMPOUNDING AND REGULATION Thus when the working point is A [Figure 5.11.(a)] the value of cos a is given by the corresponding values of V, and T,, When the point changes to B, cos a is decreased and is given by the corresponding values of V, and T,. The lower the working point, the greater will be the delay angle a and the less the value of cos a. This is also shown by. the wave diagram, Figure 5.1 1(6). The power factor introduced in the a.c. side is given by the equation: cos $=} [cos a+cos (at+y)) eee e ccc le csece ne (3.12) The less the power factor, the less will be the active power transmitted for the same alternating current and voltage and thus the greater will be the reactive power consumed woo ee ee LLL Lu d toe LL Figure 5.12. Tap changing arrangement for the converter transformers ia ihe Gotland scheme by the rectifier. Apart from this, the larger the angle, the greater will be the voltage stresses on the valves. Hence the aim should be to operate near to the top of the characteristic. Thus a lower point can be fixed, say B, so that when the working point falls, the rectifier tap changer will operate and reduce the a.c. voltage corresponding to the dotted line; in this way the working point is near to the highest level and the power factor is increased. Information can be obtained from the power factor relay on the a.c. side, so that if the power factor decreases below a certain limit the voltage will be decreased by the tap changer. The maximum limit also (say C) has to be fixed so that sufficient voltage margin is available to produce an immediate power increase if this is required by the grid control without having to resort to the comparatively slow-acting tap-change regulating arran gements. \ Sem a Mt 7 - \\ [ \ \\ [ TRANSFORMER TAP CHANGING 77 n the This upper limit is also lecessary to avoid the rectifier working on the natural voltage a and characteristic (a=0) which inplics operation of the invertor on constant current characteris- oiht, the tics. shown by hus the power factor reliy will keep the operation of the rectifier within maximum and 3 oS minimum limits by operatig the automatic on-load tap-change arrangements on each gq Be : transformer. To keep the :ower factor high, the lower limit should be brought as near -* (3.12) to the higher limit as possble. For this reason the percentage value of each tapping ghiggame should be small, but it mus: also be remembered that the tap changer should not operate on|_ med : too frequently. A range of —10 percent in | per cent or 2 per cent steps should be adequate. 3 : The operating point can iso be changed by providing a tap changer on the invertor 7 transformer. In this case the iction will be opposite to that of the rectifier tap changer. That + [ ‘ is by increasing the voltage the vorking point of the rectifier will be lifted up (high power factor) and by decreasing the voltize the working point is lowered (low power factor). But changing the invertor voltag: also means changing the power received, since it is propor- tional to the invertor voltage for constant d.c. current. Thus the current will have to be adjusted if the power is to b: kept constant. In consequence it is better to operate the rectifier tap changer. Furthemore, the voltage of the invertor should be kept high by its tap changers in order to hav: the highest operating voltage and hence the lowest copper “losses. The operation of the powe: factor relay will have to be given enough time delay (say 1 second), so that it does not sperate during disturbances of short duration. If there are no tappings left to be operate: on the rectifier, a signal can be transmitted to the invertor tap changer. The on-load tap-change arringement may be provided either on the primary or on the secondary side. Since the net axial. force binween primary and secondary windings is Proportional to the displacement between ther magnetic centres, it is desirable to maintain a perfectly balanced winding and avoid ile tap sections and extended windings for voltage control. . For large transformers, as required for h.v.d.c., it is desirable to provide a separate series auto-transformer for this purzose; the auto-transformer tap-change arrangement for the Gotland scheme is reprociced in Figure 5.12. Items 1 and 3 are two main transformers for two bridge sets in series, the secondaries being connected in delta and star, respectively, to give a twelve- Phase performance. The tectiaries of these main transformers, being equipped with on-load taps, feed th: respective regulation transformers 2 and 4, the } secondaries of which are contected in series with the secondaries of the main trans- | formers. The connections of -1e regulating windings are so arranged that they provide buck or boost Voltages suited ir phase angle for each transformer. Since the bridge sets are in suries, the two tap-changer positions need not be identical; inees | they are operated one ata time. so doubling the number of tapping steps. The automatic ‘il i } Control is so arranged that an inpulse intended for increasing the voltage affects the trans- bi former which is in the lowest pssition, and one for decreasing the voltage acts on the one in the highest position. One great advantage of the separate regulating winding is that it can be designed for low-voltaze operation. t 78 COMPOUNDING AND REGULATION Sey i 4 Jnvertor v ca a 3 | = Synchronous Load, ee edaee X E @) Jt 5.10.1 General There are three possible types of receiving systems to which the converter plant may be connected, 45) (1) When the power supplied to the receiving system is solely by means of the converter plant. This case will arise when power has to be supplied from some network to some island or isolated place where this converter becomes the only source of supply. (2) When the capacity of the receivin '§ System is very large in comparison to the converter capacity. This type will arise when power from some distant source has to be fed into a large a.c. network, or when two large a.c. networks are to be interconnected for a power transfer which is not large compared to the system capacities (i.e., the Cross-Channel project). (3) A compromise between the first two cases when the converter capacity is of con- siderabte size and comparable to the receiving system capacity. This type will arise when additional power has to be supplied to a place where some plant is already existing, which on aa RR Es REA on ee a a power S65 < & f aod gee 03 za a oa -3 REGULATION OF TIIE CONVERTER SYSTEM 719 is not very large compared to the converter plant rating, and it is intended to run both in parallel. 5.10.2 Regulation when the power supplied is only by the converter plant In this case since all the power has to be supplied by the converter system, the regulation of power, frequency and voltage has to be accomplished by this plant and its synchronous condenser. : There are three objects to be kept in mind. (1) Keeping the voltage constant at the receiving end. (2) Keeping the frequency of the a.c. system constant. (3) Running the invertor with a suitable safe margin, but not wasting the reactive power. The line diagram of this type of system is shown in Figure 5.13 (a). The power to the load is supplied solely by the invertor and its synchronous condenser. Suppose that E is the internal e.m.f. of the synchronous condenser and V its terminal : i : Vv voltage; Z, is the load impedance taking a current l= at a power factor of cos ¢. The i invertor supplics a current J; at a power factor of cos ¢; (Figure 5.13). “. I,.cos $,=I,.cos ¢ The generator current, J,=J,. sin $;,+-J,.sin ¢. and E=V+1,X, where Y, is the synchronous reactance of the generator (resistance neglected). If the load impedance Z, changes to Z’, in such a way that, for the same terminal voltage V and frequency, it would result only in an increase in the power component of the load current, giving a total current J,’’’ at a new power factor cos $’’ as shown by Figure 5.13 (0). But since the invertor current and its power factor angle remain constant (approx.), this increase in power component of current can only be supplied by the synchronous condenser. This will result in a generator power angle 8 (not shown in the figure) which will gradually reduce to zero as the gencrator slows down at a rate depending on its inertia; its frequency and hence internal generated e.m.f. will decrease until the final stable condition is reached when the load current has decreased to a value where its power component is equal to that supplied by the invertor. Under these conditions, the generated voltage will have decreased too. EF where o=/_=— New frequency _ original frequency and the terminal voltage V’=c . E-I',.¢°.X, Vv’ Riy+joe.X', components of the new load impedance Z’,. The new generator current, I’,=I,. sin ¢,+/',. sin ¢’ Also I’, cos ¢’=I; cos ¢, x, and tan ¢’=o ,— one The new load current, J,’= » where R’, and X’, are resistive and reactive sai 80 COMPOUNDING AND REGULATION Mw. Load | Converter Alternator Time Figure 5.14 Lead shaving between an alternator and a converter in parallel From these equations the new steady-state values can be calculated. Actually, due to decrease in terminal voltage (to V’), the power factor angle 4, of the invertor will increase slightly, and so will the current /, due to the sloping constant current characteristic of the rectifier; but these changes will be small and can be neglected, without affecting an under- standing of this regulation problem. It is quite obvious that to increase the frequency to the original value, the invertor current J; will have to be increased. This can be done by providing a frequency regulator on the invertor side (at the terminals of the synchronous condenser) so that when the fre- quency falls below the set value it gives a signal to the rectifier and invertor to increase their current setting and vice versa. The Steady condition will arise when the frequency returns to its set value. In the above case, the frequency and generated voltage will return to their original values when invertor current is increased to 7,"". But the invertor current has not increased to I" as is required by the load. This is so because the reactive component of invertor current and hence total generator current has also increased and the result is that the terminal voltage is still lower than the original terminal voltage and hence the active current of the load is still less than the required value. The new terminal voltage V'sE—I," X, T's 1," ve +7," . sin ” L'=> Rit j xX’, I" .cos $=," . cos 4, and tan $’=X",/R’, Thus to bring the terminal voltage back to its original value, the generated voltage will have to be increased to compensate for the additional voltage drop. This can be done by increasing the excitation of the generator. Thus a voltage regulator will also have to be provided which will increase the excitation of the generator when the terminal voltage falls below this set value and vice versa. These two regulators working together will increase the invertor current to /,’”” and generated voltage to £’”’ to meet the active load current increase. cw case = | | REGULATION OF THE CONVERTER SYSTEM 81 Lnvertor Jnverlor=——ai r I ! 1 ! ! 4 Figure 5.15 Power transmission and regulation in case of failure of communication link Similarly, it can be shown that these two regulators will adjust the converter current and generated voltage to meet the required change of reactive load current. 5.10.3 Regulation when the receiving system capacity is very large compared to the converter plant In this case the conditions are favourable to the converter plant, because the capacity of the receiving network is very large compared to that of the converter plant, and the variations of converter power have no effect on the system conditions of the network. In this case the converter plant can be set to a constant value of d.c. current, irrespective of the receiving network conditions. Constant d.c. current means approximately constant power, The converter system can also be set to a constant power by providing a power regulator. This power regulator will take into account the direct current and voltage at the invertor end. If this differs from the set value, this regulator will give the change command to the current regulators to alter their current settings. The receiving generators will take care of the frequency. and voltage of the network and there is no need for providing a frequency regulator for the converter as required in type 1. Synchronous condensers or static capacitors are not necessary if the receiving system is capable of sparjng kVAr; this is very unlikely, and it is thus desirable that the invertor end should be equipped with this supply. = = om iad & ee GT EL LT IEE IIE LE TCE OIL sean ener See anger 82 COMPOUNDING AND REGULATION 5.10.4 Regulation when converter capacity is comparable to the receiving system capacity The conditions for running a converter plant in parallel with another alternator are quite different from those when two alternators are connected in parallel.® When two alternators are connected in parallel, the allotment of active power between these alternators is determined by the speed regulation of the governors, and the magnitudes of the field currents determine the allotment of reactive power. In the case of a converter plant and an alternator in parallel the conditions are not as simple. The invertor consumes reactiye power which is approximately proportional to the active power. Suppose the synchronous condenser associated with the invertor supplies some reactive power, part of which is consumed by the invertor and part supplied to the load in parallel with an alternator. If the active power supplied by the invertor is now increased, the reactive power consumed by it will also increase; thus the reactive power supplied to the load by the synchronous condenser will also be affected. The best way to tackle this type of regulation problem is to run the converter plant as in case 1 or case 2 (5.10.2 or 5.10.3). Running it as in case 1 means causing the alternator regulators to be set to constant active and reactive powers, while the regulation of the load may be carried out completely by the converter and its synchronous condenser as in case 1. This is explained by the Figure 5.14. Suppose the load for the day fluctuates as shown in the figure. The alternator can be set at some constant power, cither as shown by the dotted line or as shown by the thick line by changing the power of the alternator according to the hour of the day, in order that converter is not overloaded or the load sharing is approximately proportional to the capacity of the plants. Another method (as in case 2) is the reverse of this, that is, setting the converter plant to constant power, with the regulation of the load left to the alternator. This method is Ly + ; Ail invertor Yy x +B (a) iy a ‘CHO Rctifian YyilR w AEs y BB = (2) if Figure 5.16 Reversing the power transmission ao na ene a en ee nan ana eel, RS TN TT [ [ ome i POWER REGULATION WITHOUT A COMMUNICATION CHANNEL 83 preferable in the sense that the alternator is already there, with all its controls, and the d.c. system is newly brought in to meet the extra demand. In this case it is much simpler to set the converter to a constant power. Another method is to evolve d.c. control equipment which has quicker action than the control equipment of the alternator, so that the control is usually done by the converter before the alternator even feels the load variations. Then, if required, the converter may be set to a constant power or constant current and the regulation wili be done by the alterna- tor: or when the converter has reached its peak value of current, the regulation will be carried out by the alternator. This is the method adopted for the Gotland scheme, where the converter plant has to work in parallel with the Gotland steam plant.@” 5.11 Power transmission and regulation in case of failure of the communication channel Since normal operation and control of regulation of power transmission is carried out by giving change commands to the rectifiers from the invertor end, it is desirable to ensure power transmission in the event of failure of the communication system. It is comparatively easy to carry out transmission without changing the current setting (without power regulation). It is preferable, however, that it should be possible to control the power regulation entirely from the invertor side; this is particularly important if the h.y.d.c. link is the only source of power. This is possible, as indicated in Section 5.8, by providing a current regulator on the invertor. The continuous lines in Figure 5.15 show the normal working of the system, the current corresponding to point A. When the communication channel fails, the invertor voltage can be automatically in- creased by the tap changers so that it works on the constant current characteristic whilst the rectifier works with no delay, as shown by the dotted line. The current now corres- ponds to point B. The constant current regulator of the rectifier can also be given its maximum setting. The regulation can now be carried out by changing the current setting of the invertor regulator. Extra consumption of reactive power will arise with this method of operation, although automatic tap changing will minimise it. 5.12 Stability of the receiving system The stability of the receiving system depends not only on the d.c. link but also on any synchronous generators and a.c. system which may be in parallel with the d.c. link. The simplest case is clearly the one where the d.c. link is the only source of power to the receiving system. The frequency of this latter system will clearly fall if the d.c. supply is interrupted for any length of time. If the control circuits of the d.c. system are so designed that they can operate satisfactorily at a low frequency, the a.c. receiving system should be able to establish itself once the d.c. supply is restored and bring the frequency back to normal. The grid control gear at Gotland “°9 has been designed to function down to 25 c/s, and if the interruption has not been prolonged operation can resume and the system accelerated to 50 c/s. A further point is that by tripping the circuit breakers on the a.c. side and de-exciting the synchronous condenser field, the deceleration of the machine can a attain nn tN A RE Rn MR a ee a = gegen See Revie eernmnjememy totter cine -roRTS Spe eNaREOn rnenmeeyerF7sHTSTORP-TPEELTEIREP gm YI rharmenenenareNnens te yarennstnir of cab ice SRLS i a 84 COMPOUNDING AND REGULATION be so delayed that the system can be started up in the manner above after an interruption i as long as ten minutes. ; 5.13 Reversing the direction of power transmission It has already been demonstrated that a rectifier can be made to operate as an invertor by providing a phase advance of more than 90° and rather less than 180°; the reverse is clearly true. Figure 5.16 shows the h.v.d.c. system in which A is the rectifier and B the invertor. Now if the angle a of the rectifier A is. advanced by more than 90° and the angle f of the invertor Bis advanced by more than 90°, then A will become the invertor and B the rectifier. Current [" will still be flowing in the same direction, while the voltage sign of the d.c. line will change . | giving a power transmission from B to A, as shown in Figure 5.16 (b). i Since the firing angle can be changed automatically as required, it is possible to carry out power reversal automatically as dictated by the a.c. systems at each end of the d.c. link, Although stability problems of the controllers arise, they are not insuperable. The straightforward interconnection of two a.c. systems is relatively easy by this method. Changing the polarity of the d.e. line is not, however, suitable if there is the possibility of interconnections on the d.c. side or when there is a tapped line on the d.c. side. In such a case power reversal can be carried out by reversing the terminals of the station, thereby maintaining the same line polarity. , Reversal of polarity may also be undesirable from the standpoint of insulation, parti- cularly in the case of cables. 7 ; é £ a ie Bk: try Pei lity gy A CHAPTER 6 Protection of H.V. D.C. Systems References: (3) (6) (9) (10) (30) (36) (49) to (61) inclusive, (104), (105), (109), (112), (216). 6.1 Basis of protection of the a.c. and d.c. systems Adequate protection of all plant, cables and transmission lines on a power system is a first essential and is based, in the a.c. case, on the provision of circuit-breakers at critical points; the circuit-breakers are controlled by relaying arrangements which operate the switches within specified time limits and with suitable discrimination. On high voltage systems, the provision of high-speed auto-reclosing circuit-breakers may be necessary as a means of maintaining continuity and avoiding synchronous instability. The circuit-breaker is thus the basis of protection in the a.c. system and can be so because of the occurrence of current zero twice every cycle; from ari enginecring standpoint, the problem is one of quick mechanical operation, together with rapid deionisation of the are by thermal and/or electro- magnetic means to avoid restriking. Since there is no current zero in the case of d.c., the circuit can be broken only by extending the are sufficiently to bring the current to zero. Air blast circuit-breakers can be developed on this basis up to a certain size, but when the rating dictated by a given h.v.d.c. project is considered, the length of the are to be drawn is so great that it becomes impracticable. Other methods discussed later (Section 6.10) also fail to provide a practical solution, although research in the U.S.S.R. seems to be leading to promising results.@10 Fortunately the converting equipment possesses the features that protection is possible by grid control; furthermore, the specd of operation is somewhat greater than that of circuit-breakers. Complete protection could otherwise be provided by installing a.c. circuit-breakers on the a.c. side of the rectifier-transformer, but this would involve inter- ruption on the a.c. side for every type of fault. Difficulties can be foreseen for the cases of interconnection of d.c. lines and tapping from a main d.c. line; it is then that the necessity for d.c. circuit-breakers arises. But for trunk linc transmission (isolated d.c. line), protection by grid control (backed up by a.c. circuit- breakers) is adequate and efficient. Many types of fault can occur in the various sections of h.v.d.c. systems, either on the a.c. or d.c. sides; and proper measures must be taken to deal with each of them. It is advisable to consider each fault separately and to study its effect on the system before deciding the system of protection. The main aim should be to have as little interruption as possible but not at the expense of putting a heavy duty on valves and running a risk of damage. If interruption is necessary it should be as short as possible to minimise the inconvenience and loss of revenue involved. The methods available for protection and the severity of faults are very dependent on the method of regulation and compounding provided. Constant current regulation will not allow the fault current to rise above the previously set value; proper compounding of invertors will not allow commutation failures to occur in the case of, say, an instantaneous reduction of the alternating voltage if the d.c. smoothing reactor is large enough to diminish the current derivative to a large but definite value. Sombie 86 PROTECTION OF H.V.D.C. SYSTEMS 6.2. By-pass valve 6.2.1 General A by-pass valve is an essential part of the protection arrangements. This is a grid con- trolled valve connected across the bridge circuit, as shown in Figure 6.1. This valve is kept blocked by a negative voltage on its grid, when the bridge unit is conducting in the normal manner. Due to various faults (as discussed in later sections) it may be necessary to stop the bridge unit from conducting. This is done by stopping all the grid pulses on the valves. If there is another bridge unit in series, its current should be allowed to pass. Figure 6.1 By-pass valve in a bridge unit This is done by opening the by-pass valve of the first bridge as soon as that bridge unit is blocked. When the fault is cleared, the bridge is unblocked, and the by-pass valve is blocked. The manner of operation of the by-pass valve has to be considered in the light of the possible system arrangements. 6.2.2 By-pass valve operation in a system using one bridge unit at each end Figure 6.2 (a) shows this simple system with both the rectifier and invertor bridges equipped with by-pass valves. Considering firstly the operation of the invertor as indicated by the waveform of Figure 6.2 (b); suppose that invertor blocking takes place at instant A, i.e., no further valves are fired, but the by-pass valve is still blocked. Valves 5 and 6 will continue conducting; the invertor voltage will decrease, and result in increased current on the d.c. side. ; After point C, the invertor voltage reverses and the invertor becomes a rectifier, thereby resulting in a further increase of current. Although the rectifier will control the current, the blocking results in a short-circuit on the d.c. side and the two valves which were con- ducting at the instant of blocking will not stop conducting until the rectifier, i.e., the other end of the system, is also blocked. Now suppose that the by-pass valve is opened, along with the blocking of bridge valves at instant A; the by-pass valve will short-circuit the d.c. and it will take over current from the bridge valves thereby. completely relieving the bridge unit. Consider now the rectifier side. When current is flowing in the system, there is a large amount of energy stored therein (mainly in the smoothing inductor). Suppose that the valves are blocked at instant 4 {as shown in Figure 6.2 (c)] when valves 2 and 3 afé conducting, subsequent to a line fault. Although no further firing of valves will take place, valves 2 and 3 will continue conducting; after point C, the voltage becomes See De eke et i, a OR RE IOC STE peep re : a a mnt s BY-PASS VALVE OPERATION 87 em em mT om! ‘9 @ th ary ian oi" vy = | LAHEY 6 a 4 6 2 © Figure 6.2 (a) (6) (c) By-pass valves in a system having one bridge unit on each side 5 o negative and valves 2 and 3 should then stop conducting, but it may well be that this will not take place, since the stored energy in the system will continue to cause current to flow and overcome the negative voltage. If the stored energy is large enough to continue the flow of current until point D, then blocking is no longer possible since the voltage between Y and B becomes positive again after D. If, at point A, the by-pass valve is also opened, then up to point C it will not conduct since its cathode is positive with respect to its anode. After point C the by-pass valve will take over current from valves 3 and 2, and the bridge circuit will be blocked. 7 The by-pass valve will continue to conduct until the stored energy is dissipated. After point C, the transfer of current will take some finite time for commutation which depends on the are voltage and transformer Icakage reactance. % “i my oa 4 come © aan via r —— = nian ia nin lm ti 88 PROTECTION OF H.V.D.C. SYSTEMS It is clear that a by-pass valve is an essential requirement for cach bridge. After blocking the bridge valves and unblocking the by-pass valve, it is necessary to start the bridge circuit and block the by-pass valve; this is considered in Section 6.2.3. 6.2.3. By-pass valve operation in double-bridge connection, both bridges on the same side of earth Suppose that bridge-unit A on the rectifier side (Figure 6.3) is blocked and its by-pass valve opened. The flow of current will be as follows: B-by-pass valve of A—transmission line-+C->D->earth connection. It might be considered that a single rectifier unit such as B could not operate in a sustained fashion against two bridge units at the invertor end of the system; with suitable constant current compounding of the invertor (as explained in Section 5.8), the back voltage of C and D will be reduced by the firing angle 8 however, and thus the current will be maintained constant at the pre-set value, although the received power will be reduced by half. The invertor will run at a large angle for a short period only, since in most cases the recti- fier will be automatically unblocked a few cycles after the initial blocking. If the rectifier bridge is blocked for a longer time, then an invertor bridge will be blocked automatically after working with a large angle for a few cycles. If the invertor is not provided with a constant current regulator, then an invertor bridge will also have to be blocked whenever a rectifier bridge is blocked, otherwise the system will run down. Similarly, in the event of invertor bridge C being blocked and its by-pass valve opened, the constant current regulator of the rectifier will advance the angles of A and B so that constant current is maintained. Figure 6.3 By-pass valve in a double bridge system having both bridges on the same side of earth Once a by-pass valve has fired it can be blocked only by first interrupting its current so that its grid can regain control. In the case of rectifiers this is done automatically since, once the bridge valves are started, the bridge unit establishes a positive voltage across the by-pass valve, the cathode of which becomes positive with respect to its anode. Bridge valves then take over current from the by-pass valve, which stops conducting since it cannot conduct in the reverse direction. BY-PASS VALVE OPERATION 89 In the case of the invertor, suppose that unit C is blocked and its by-pass valve is conduct- ing; if unit C is now started, it cannot continue in operation since the by-pass valve will not stop until its cathode is made positive with respect to its anode. The necessary reversal of polarity may be accomplished by increasing the angle B to in ddeaf greater than 60°, since for B>60°, the voltage reverses for the periods in excess of 60° for =] every occasion of firing. This excess angle should be enough for commutation of the current “pass ; and deionisation of the valve. This arrangement ensures stopping of the by-pass valve } and the taking over of the current by the bridge unit. As soon as the current transfer is f ; complete, the angle 8 will be reduced to its normal value. ined slaps i 6.2.4. By-pass valve operation in double-bridge connection, with earthed neutral : [ . In this case, the flow of current I, is as shown by the full arrows of Figure 6.4, i.e., from rectifier unit A, through the positive line, through invertor C and then to earth; rectifier B forces current J, through the earth, invertor D and the negative line as shown by the dotted fF arrows in the same Figure 6.4 iff In steady-state operation the currents into the earth connection should cancel each other, vally but this can only be attained if the conditions of operation for the individual bridge sets on fou each side of earth ensure equal currents. Thus if the voltage of invertor C drops, it is af necessary that the angle of rectifier A only should be advanced, thereby reducing the voltage tem ‘ of A accordingly in order to maintain the same constant current T,in both conductors. It vags . i is necessary to provide two Separate current regulators, working on the positive and negative i i side of earth, for A and B respectively; both regulators must be given equal current settings. The same applics to the invertor side; only one regulator will be required for two or more : bridges in series on the same side of earth. : [ Suppose now that invertor C is blocked and its by-pass valve opened; the regulator of 4 3 will advance its angle to a value somewhat less than 90° and reduce its output voltage so as to maintain the current J, through the positive conductor. Since there will be a short period of time before full operation of the regulator, there may well be a transient condition of current through the earth. Rectifier bridge Inverton bridge units é units Figure 6.4 By-pass valve in a double bridge system with earthed neutral Bi ka) 90 PROTECTION OF H.V.D.C. SYSTEMS Suppose that rectifier A is blocked and its by-pass valve opened. The current regulator of C will advance its angle 8 by more than 90°, causing C to work as a rectifier and maintain a current /, through the positive conductor. If the current regulator is not provided on the invertor side, or if invertor C is not allowed to advance more than 90°, then the positive side of the system will run down and the negative side will conduct through the earth connection as shown by the dotted arrows of Figure 6.4. If it is required to shut down the bridge on one side of earth, then the current through the carth will have to be tolerated. Unblocking of the bridge units will be as described in Scction 6.2.3. j These considerations show that the d.c. system might well be earthed at one point only, either on the rectifier or on the invertor side; in this case one common regulator could be used, ae 3 + 4 othen bridges Figure 6.5 Disconnecting a bridge unit 6.2.5 Disconnection of a bridge unit ; In a system employing more than one bridge unit at each end, it may be necessary to disconnect and remove a bridge unit for servicing without disturbing the operation of the other bridge units. This can be accomplished by providing isolating switches as shown in Figure 6.5. First, the bridge to be disconnected is blocked and its by-pass valve unblocked to take over the current. Switch 1 is then closed, thus short-circuiting the by-pass valve and taking over the current from it. Isolators 2, 3 and 4 may then be opened to separate the bridge unit from the a.c. and d.c. terminals. The bridge unit should be earthed. Although desirable, it is not possible in practice to arrange for the bridge to be withdrawable as a complete entity because of the dimensions and weight of each valve unit. In order to reconnect the bridge into the system, the reverse procedure should be followed. Isolators 2, 3 and 4 are closed, and this is followed by the unblocking of the by-pass valve and the opening of switch 1. The current will transfer from the switch to the by-pass valve; the voltage appearing across the switch 1, when it is opening, is the voltage drop in the by-pass valve and hence is a very low-voltage switch. The bridge is then unblocked and takes over the current from the by-pass valve, which is blocked; however, before unblocking the bridge, it should be made certain that complete current Transfer has taken place from the short-circuiting switch 1 to the by-pass valve, otherwise a short-circuit across that bridge will result. na cn a ee ct ti ne G G nS a as sa z rt. ws CLASSIFICATION OF FAULTS 91 6.3 Classification of faults The faults can be classified as follows: 6.4 Faults on the a.c. side of the invertor. 6.5 Faults on the invertor plant. 6.6 Faults on the d.c. line. 6.7 Faults on the rectifier plant. 6.8 Faults on the a.c. side of the rectifier. 6.4 Faults on the a.c. side of the invertor All these faults will result in reduction, distortion or collapse of the a.c. voltage on the invertor terminals. These faults can be classified into distant faults and near faults. . Receiving system Inverton A JY) Syrichronous condenser (2) Receiving system / Ihverlon KD). A Synchronous condensen (b) Figure 6.6 (a) (6) Faults on the a.c. side of the invertor . | | aig aan inane issih SSE aS HMR SIs emer one oa 92 PROTECTION OF H.V.D.C. SYSTEMS 6.4.1 Distant faults Distant faults are not necessarily geographically distant but are electrically distant in the sense that they are beyond the main transformer at, say, point A in Figure 6.6. Due toa fault at point A, or further away, there will be some reduction in the voltage at the invertor terminals. The voltage reduction depends upon the transformer short-circuit capacity and the reactance between primary and secondary. A three-phase fault will result only in reduction of voltage; while two-phase and single phase faults will result in reduction as well as distortion of voltage. This reduction in invertor a.c. voltage can have the following consequences: (i) Arise in the value of the d.c., which may be prevented by the use of rectifier constant current compounding, although a transient rise in current cannot be avoided. (ii) Increase in commutation angle, which may result in commutation failure. The commutation angle will increase even if the d.c. is constant since for constant current, the commutation angle is inversely proportional to the commutation voltage. (iii) Due to distortion in voltage in the cases of two-phase and single-phase faults, the available angle for commutation will increase for two valves and decrease for an- other two valves. With decreased angle there may be failure to commutate, Figure i There is thus a general tendency towards commutation failure, and it is clear that special provision should be made (because of the high incidence of these faults) to prevent failure in invertor operation as a result. 7 Commutation failure can be prevented by making the angle B sufficiently large for normal operation of the invertor, so that when a voltage reduction, of say, 20 per cent, occurs there is sufficient margin for safe commutation at the reduced voltage. This means pro- viding sufficient reactive power to provide sufficient angle 8, during normal operation.“ Another method is to provide a suitable compounding of the invertor as explained in Chapter 5, so that under all reductions and distortions of a.c. voltages, the angle £ is auto- matically adjusted to a safe value, , ' To reduce the fluctuations due to this kind of fault, it has been suggested that the syn- chronous condenser should be connected to the tertiary winding of the converter trans- former 6 as shown in Figure 6.6 (b) and the tertiary winding designed in such a way that its leakage reactance is small towards the invertor and large towards the system side. This method maintains the voltage on the invertor terminals. Ifa synchronous/static capacitor combination is to be provided, the synchronous capacitor should be of sufficiently large rating compared to the static capacitor in order to maintain as high a voltage as possible at the invertor terminals during faults on the a.c. side. Also the larger the size of the reactor on the invertor d.c. side, the more improved will this situation become on account of the increase in time constant of the line discharge current.’ 6.4.2 Near faults A fault at B, Figure 6.6, implies complete collapse of the a.c. voltage on the invertor and the short-circuiting of the d.c. at the invertor end. The rectifier will prevent long-term rise in current, but due to line capacitance, there might be considerable initial value of current. trent, [- r an- igure 1 lure INVERTOR FAULTS 93 The path of this particular short-circuit will first of all be through the phase windings; and if firing continues it will result in a direct short-circuit to a pair of valves. It is likely that proper compounding of the invertor will enable it to recover by itself as soon as the a.c. voltage is established; if this is so, blocking of the invertor will not be necessary. This type of fault can be treated as a commutation failure, as in the next section. 6.5 Invertor faults 6.5.1 Commutation failure The most severe invertor fault is commutation failure. With correct compounding of the invertor these faults are very greatly reduced, but still commutation failure may be unavoidable if the reduction in a.c. voltage takes place during the commutation process, or excitation failure occurs. SS) @® Invertor back voltage ‘ during single commutation failure iv (o)Jnvertor back voltage during double commutation failure Figure 6.7 (a) (6) (c) Commutation failure 94 PROTECTION OF H.V.D.C. SYSTEMS A is the point where valve 3 fires (Figure 6.7) and commutation from’l to 3 is expected to take place. Suppose the commutation is not able to take place or valve 1 has not deionised until point B. After B, the anode voltage of 3 becomes negative with respect to that of 1; hence commutation is no longer possible, valve 1 conducts again and valve 3 stops. Valve 1 continues to conduct together with valve 2. The back voltage reduces exactly as the voltage between phases R and B after point B. At point C valve 4 fires. Commutation takes place from 2 to 4 [Figure 6.7 (a) and (6)]._ This makes valves 1 and 4 conduct together and d.c. short-circuit results. It is quite clear that the short circuit current does not pass through the transformer wind- ing after this. At point E nothing happens when S fires, since the anode voltage of 1, which is conducting is positive with respect to that of 5. At point F the current commutates from 4 to 6. At this point the back voltage becomes negative for a short while, until G, beyond which the back voltage establishes itself and the short-circuit is over. At H the commutation of the same valve takes place once again, but this valve is expected to recover in the meantime and normal operation is established without much disturbance. Between points B and C the current passes through 1, phase R, phase B and 2, and during this period there is some rise ini current, though this is not great because of transformer inductance and the smoothing choke. It is, however, possible that it may rise sufficiently to cause a commutation failure of valve 4 from 2 [Figure 6.7 (a) and (c)]. This results in still more severe conditions, since it can be seen that after point D, the voltage between phases B and R reverses, and hence this a.c. voltage instead of being a back voltage, adds up with the d.c. to give a severe short circuit of the d.c. and a.c. voltages through the trans- former windings. The invertor becomes a rectifier in series with the rectifier. At point £ nothing happens since the anode voltage of 5 is negative with respect to 1, which is conducting. At F, nothing happens for the same reason. At J, nothing happens since 1 is already conducting. At J, the a.c. back voltage becomes positive and establishes itself thereafter. At point H, commutation of the same valve takes place and normal . operation may be resumed although it is now less likely due to the substantial rise in current and ionisation of the valve. In the case of double commutation failure, it can be seen that part of the short-circuit d.c. passes through a pair of valves, by-passing the transformer windings for an interval CD’, which is the commutation time from 2 to 4 and back. With correct invertor com- pounding there is no reason why successful commutation should not take place at the next firing (valve 2 to 4 in this case) so making a double commutation failure very unlikely. If it should occur by chance, the invertor may even recover from sucha fault. in the event of it occurring and being of a permanent nature, the only remedy is blocking of the inverter valves and opening of the by-pass valve. A suitable method of protection is to use a relay which compares the currents on the a.c. and d.c. sides, the former being rectified. As explained above, the short-circuit current by-passes the a.c. windings for 120° in a single commutation failure and for the interval CD’ in double commutation failure; thus for every commutation failure, the d.c. will exceed the a.c. An arrangement can be introduced such that if this happens for two suc- cessive-cycles, the invertor will be blocked. This is clearly a good arrangement since it provides for the eventuality of automatic recovery of the invertor. INVERTOR FAULTS 95 A further method is to measure the voltage across the invertor bridge, or across each bridge if there is more than one. If the voltage becomes zero for approximately 120°, it means a single commutation failure; if it becomes negative for approximately 120°, a double commutation failure is implied. Blocking of the invertor bridge will, of course, be accompanied by the opening of its by-pass valve and followed by unblocking (as explained in 6.2.3) after about five cycles. This short time of interruption will have little effect on the receiving a.c. system. 6.5.2. Fire-through or grid blocking failure This type of failure may also be very severe and as bad as commutation failure. Consider the instant P [Figure 6.7 (a)]. At this point the anode of the valve 3 is positive, but it is held from conducting by the negative bias on the grid. Suppose at point P its grid fails to hold the valve from conducting and valve 3 fires. Since valve 6 is already conducting this implies a d.c. short-circuit. At instant Q, when 2 fires, there are two possibilities: (i) Commutation takes place from 6 to 2. In this case, after R, the back voltage will build up. At C commutation from 2 to 4 takes place. At E commutation from 3 to 5 will start and if grid failure is temporary, the commutation will take place; if permanent, it will result in commutation failure. (ii) If the commutation from 6 to 2 does not take place successfully or if the grid failure occurs between Q and R the d.c. short-circuit will continue until Z. At E valve 5 fires and if the fault is not permanent, commutation from 3 to 5 will take place and back voltage will be built up and the converter will recover by itself. If the fault remains, the d.c. short-circuit will continue and the invertor will be blocked and then unblocked after a few cycles. If the fault is permanent, the short-circuit will be repeated and the invertor will be blocked again. In that case, it may be necessary to effect a shut-down to change the valve. An automatic arrangement can also be made to detect the faulty valve if the short-circuit is repeated. This type of fault may be caused by the failure of grid bias, the effect of voltage surges on the valve or defect in the grid of the valve. 6.5.3 Arc quenching and failure of a valve to fire Failure of a valve to fire can occur as a result of pulse failure or excitation failure, the final result being commutation failure. Arc quenching can arise on account of the high current surges which are likely at the instant of valve firing (further discussed in 6.13.3), or because the valve temperature is too low (Chapter 13). If arc quenching takes place the valve cannot pick up current and the result is commutation failure. 6.5.4 Invertor backfire Backfiring occurs when anode voltage is negative with respect to its cathode and the valve conducts current in the opposite direction (from cathode to anode). On invertor valves the anode voltage is negative with respect to cathode only for a short period ab, as shown in the valve voltage diagram (Figure 3.12). This period is very short and the voltage is too small to cause any backfire. ET my ahah u ened i a ; = E a 96 PROTECTION OF H.V.D.C. SYSTEMS But sometimes, when the regulation is being controlled by the invertor, the angle B may be large and a backfire might occur, though the probability is very small. If the back- fire occurs, it will clearly result in short-circuit of two transformer phases and the short-circuit current will be limited only by the leakage reactance of the windings. Even though the voltage changes sign at 5, due to the inductance of the transformer winding, current will come to zero at point ¢ and this means commutation failure. For the remainder of the cycle, the conditions will be the same as during commutation failure. If the reason for the backfire is temporary, the invertor may recover; if the faulty valve needs considerable time to deionise, the commutation failure may well be repeated. This fault differs from commutation failure in that, apart from the short-circuit of d.c., an a.c. short-circuit also takes place for a short interval, ac, of two transformer windings independently of d.c. This fault may, however, be treated as a commutation failure in the sense that blocking will take place if backfire, and hence commutation failure, repeats itself. Since this parti- cular fault in invertors is very unlikely, this protection should be adequate. But if it is desired not to allow this backfire to be repeated and the invertor be blocked, as soon as the first backfire takes place, it will be necessary to discriminate between the two faults—com- mutation failure and backfire. This may be accomplished by the use of a comparator relay (as mentioned in Section 6.5.1) for the detection of commutation failure. The com- parison is of current from the d.c. current transformers on the d.c. line, and the rectified current from the a.c. side. If the current on the a.c. side exceeds that from the d.c. side, then a backfire is indicated and the invertor can be blocked. 6.5.5 General observations It is clear from the above that every type of invertor fault results in commutation failure. If every effort is made to ensure that the next commutation after the commutation failure takes place correctly, then the invertor will generally recover automatically without any blocking and the effects of fault will hardly be felt on the receiving a.c. system; most faults will be cleared in this way and the necessity for blocking will thus be rare. To effect block- ing only after the repetition of fault, should not create severe stresses on the valves since fast-acting constant-current compounding of the rectifier will not permit of current rise except that due to discharge of line capacitance. Some stress is, however, unavoidable, and the engincering arrangements should be such as to minimise the occasions when block- ing action is necessary, 6.6 Faults on the d.c. transmission line This type of fault results in heavy current in the rectifier, if it is not compounded. But due to the constant current compounding of the rectifier, the rectifier angle a will increase and it will work at somewhat less than 90°. This fault will also result in stoppage of the invertor since the d.c. voltage is cut off, and should be cleared as early as possible. This can be done only by blocking the rectifier. For example, suppose an insulator has flashed over due to some temporary reason, the arc due to flashoyer, once started, will not be extinguished by itself until the current is stopped. Hence this needs blocking of the rectifier followed by unblocking, as soon as the current falls to zero. Ne Fon Saianniemme ee fener 7 TT ~— ng a me ” ied met rg Un mee F bao oO" RECTIFIER FAULTS 97 If constant current compounding is provided at the invertor as well (as discussed in Chapter 5), the invertor angle will advance to more than 90° (if allowed to do so by com- pounding) and work as a rectifier to maintain the current as fixed by its current setting. In this case, the current through the fault will be only that corresponding to the “margin setting”, i.e., the excess of rectifier setting over the invertor setting, there being a much greater chance of quick recovery and less chance of permanent damage caused by the fault arcing. Nevertheless, rectifier blocking is necessary to clear the fault; if there are a number of bridges in series on the same side of earth, then it is necessary to block all of them. If, however, the system uses two outer conductors with earthed neutral point, then it is only necessary to block the side (positive or negative) which is faulty and the supply through the other side can continue uninterrupted with the earth as the return conductor. If the fault is of a permanent type then restarting will not help, and the rectifier will be blocked again. In that case a shut-down until the fault is repaired, or a changeover to another conductor (if available) is necessary. These faults can occur due to a broken or contaminated insulator, icing on the insulators, overvoltages and, in the case of cables, failure of the cable insulation. Submarine cables are also susceptible to mechanical damage caused by shipping activities. For such cases, the only remedy is to block the rectifier; care must, however, be taken to ensure that the rectifier is not blocked as a result of an invertor fault (which also appears as a d.c. short- circuit when viewed from the rectifier), since the invertor is capable of recovering from its own faults. Discrimination is thus necessary between invertor faults and d.c. line faults, and the latter can be detected by measuring the voltage on the d.c. output side beyond the rectifier smoothing reactor since a fall to near zero will occur for a d.c. line fault. A delay of up to half a second could additionally be given, so that the invertor is afforded time to recover. An indication of d.c. short-circuit can also be obtained by measuring the angle of firing of the rectifier, or the power factor on the a.c. side. If the firing angle is advanced beyond certain limits for more than half a second, then the rectifier will be blocked. If it is not desirable to give a time delay then rapid discrimination is necessary between the invertor and the d.c. line faults. This is possible by measuring the rate of fall Il of voltage; this will be much greater in the case ofa d.c. line fault than of an invertor fault since the latter implics a d.c, fault beyond two reactors whereas the former fault is beyond one reactor only. Rapid discrimination may also be obtained by sheasuring rate of rise of direct current, and may be accomplished by measuring the transient voltage across the rectifier smoothing inductor by means of an instrument transformer.5) Discrimination between the two locations of fault may be obtained by setting a limit of di/dt due to invertor faults below which the rectifier will not be blocked. Arrangements may also be made for automatic reclosing after a predetermined time, as in high voltage, a.c. transmission line practice; if the fault is of a temporary nature, trans- mission will continue after a short interruption. 6.7 Rectifier faults) 6.7.1 Backfire or arc-back®2) This is the’loss of the unidirectional conducting property of the valve, and the valve ———, — a seit acc — Current in valve | (C) Current in valve 5 Figure 6.8 (a) (6) (c) (d)_ Backfire | in a rectifier bridge unit een rennet FR RETO an u cr €4 woo RECTIFIER FAULTS 99 fires in the opposite direction; its cathode is positive with respect to its anode and current flows from cathode to anode. This is a very severe type of fault and causes a heavy current and virtual short-circuit between phases on the a.c. side. Backfires are random phenomenon. Various reasons and explanations have been offered to explain the causes of arc-back, and they may be reduced by appropriate precautions. But there seems to be no limit beyond which the arc-back may be said to occur, and within which they may be said not to occur. Referring to Figure 6.8 (b), arc-backs will be most likely to occur at A, Band C; at A and C due to sudden changes and oscillations caused by sudden changes in voltages (discussed later in 6.12 and Chapter 12) or at the peak value of the back-voltage at B. Suppose that valve 5 backfires, say at point A, as in Figure 6.8. At that moment the valves 1 and 6 are conducting in the normal way through phases Rand Y. When valve 5 backfires, it will short-circuit phases R and B seriously. The current will rise in valves 1 and 5 as shown, and the rate of rise will be limited by the leakage reactance of the trans- former windings. Hence to avoid a large short-circuit current, the leakage reactance of the transformer windings should be fairly large. Duce to collapse of the rectifier voltage on the d.c. side, the current in the line will gradually fall to zero. It should be noted that rectifier backfire is considerably more severe than invertor backfire since on the rectifier valves the negative voltage exists for a large part of cach cycle, about 240°, compared to a very small angle in the case of the invertor. Short- circuit currents may attain values greater than ten times the normal full-load current, and rectifier blocking is then essential. Due to the large amount of inductance in the circuit compared with the resistance, the short-circuit current between phases B and R will be at its peak at the voltage zero D, assuming that no further firing takes place after backfire, i.e., valve 3 does not fire. If valve 3 does fire, the short-circuit current will commutate from valve 1 to valve 3 and the peak of short-circuit current will then occur at E; speed of blocking is clearly of great importance. Furthermore, in spite of blocking, it is possible that because of the heavy current, valve 1 has not been completely deionised up to point F and valve 5 has lost its valve action, i.e., it is ready to backfire again. In this case the short-circuit will be repeated and the only protection is tripping of the a.c. circuit-breaker; this will be a rare occurrence. As mentioned before, the most rapid way of detecting rectifier faults is by means of a relay comparator which compares rectified a.c. input with d.c. output; with the former excecding the latter, the rectifier is then blocked. Blocking should be followed by unblocking after about five cycles, so that the supply is interrupted only for a time sufficient for the faulted valve to recover. If the fault is per- manent the bridge will be blocked again, and arrangements invoked for taking it out of Service. The frequency of arc-backs can be reduced by proper valve design (Chapter 12), and by the damping of plant and circuit oscillations (Section 6.12). 6.7.2 Failure of a valve to fireSY This constitutes failure by the valve anode to pick up the arc. Suppose at A, when valve — jabato mie amount of a.c. at 50 c/s on the d.c. side ; it will also cause d.c. magnetisation of phases Rand Y, since phase R is conducting for muck i more than 120° in the positive direction and phase Y phase B as well, but only sniall, whilst the Positive mag greater magnitude. This type of fault does not do any major damage limbs, and some method should be adopted to detect t the valve. This fault can be detected by measuring the voltage on the d.c, output side, before the choke, d.c. Another method is to measure the apart from d.c. magnetisation of the his fault if it continues, and to replace the alternating current component of or the alternating components of the positive and negative currents in the transformer wind- ication can be obtained as to which valve has failed. current. Faults of this type can be caused by excitation failure or failure of the associated circuits to produce pulses correctly, 6.7.3 Fire-through This fault can be caused by failure of the grid bias. In this case the valve will start conducting just after its anode voltage becomes positive wi I cause a slight d.c. magnetisation of the limbs. The effect of d.c. magnetisation will be that one phase will be magnetised slightly in the positive direction and the other in the negative direction; an indication can easily be obtained from the method explained above (Section 6.7.2). BS RI ¥3 B5 RI 6 2 4 6 7) Figure 6,9 Failure of valve firing in a bridge rectifier ta alt ee Sesiittened ete ie ani el niin ine stn [ FAULTS ON THE A.C, SIDE OF THE RECTIFIER 101 in 6.8 Faults on a.c. side of the rectifier aL These faults may be either distant or near faults. They will result in either reduction or able collapse of the a.c. voltage. In the case of reduction, proper invertor compounding (Chapter a ; 5) will prevent the system from running down. In the case of collapse the supply must be ce interrupted. ion, ‘An a.c. circuit-breaker must be provided to interrupt the circuit in case of transformer eof faults or when rectifier grid control fails to interrupt the backfire. This circuit-breaker is ’ r necessary in the normal way to connect or disconnect the d.c. transmission system. 6.9 Failure of supply to auxiliary equipment If the supply to the auxiliary equipment is from the a.c. system, a fault on the a.c. side will result in the supply being cut off from the auxiliaries as well. It is essential that the supply to the valve auxiliaries should not be interrupted, especially to the exciter anodes and the grid bias. Thus it is important to provide for these auxiliaries separately, e.g., by a diesel generator. . On the invertor side, this generator can also be used to bring the synchronous condenser to its full speed before connection to the invertor. This last provision is clearly not neces- sary if the a.c. supply on the invertor side is already available; the synchronous condenser could also be started directly from the d.c. side. 6.10 D.C. circuit-breakers 6.10.1 The need for circuit-breakers There is no immediate demand for circuit-breakers for h.v.d.c. systems since the schemes at present existing, under construction, or contemplated, are trunk systems. For any such system, interruption by means of grid-blocking signals is efficient and effective provided that back-up protection is provided by means of circuit-breakers on the a.c. side of the rectificr banks. The question of interconnection of d.c. transmission lines may, however, arise in the L future and it is appropriate that this chapter should present some discussion of the problems which could then occur, particularly since d.c. circuit-breakers would be needed.(112) Suppose AOB is the main d.c. line, as shown in Figure 6.10 (a), and OCisatic line. If [ there is a fault on line OC, the rectifier A can deal with it by grid blocking, but it means complete shut-down for a fault on the tapped line. When the fault occurs on OC, constant current compounding of the rectifier will not 7 : allow an increase in the total current which it supplies, but the effect will be a rise of the : short-circuit current in OC accompanicd by a reduction of the current in OB; finally the whole of the current will be transferred to OC. : Hence it is necessary that OC should be disconnected as quickly as possible, and the [ supply to the receiving system of B restored so as not to affect its stability on the a.c. side. In such a case a circuit-breaker at the tapping point would be necessary. A further case could arise if transmission was to be carried out by means of a number of : parallel circuits as shown in Figure 6.10(b). Due to the non-availability of d.c. circuit- [ breakers, interruption would have to be by means of rectifier blocking action followed by isolation of the faulty line. This is clearly not such a difficult problem as that of the tapped line, ; - a rete a ~ Hed 3 102 PROTECTION OF H.V.D.C. SYSTEMS With a.c., the current comes to zero every half-cycle and advantage is taken of this in order to break the circuit, and the problem for the circuit-breaker becomes the prevention of restriking. With d.c. no such current zero exists and the current has to be brought to zero (from the value of short-circuit current) by some means. This implies dissipation of the energy stored in the circuit inductance at the beginning of interruption, plus the energy supplied by the rectifier during the time of interruption. Rectifier = (b) a Figure 6.19 (a) (b) Cases of d.c. transmission where circuit-breakers could be necessary 6.10.2 Possible types of d.c. circuit-breakers This problem has attracted attention for a number of years past have been advanced. Some are included here: (a) By inserting resistances in the circuit to be disconnected duri tion; this absorbs energy and limits the fault current to th The method is not practicable for h.v.d.c., on account of the , and many proposals ng the switching opera- ¢ lowest possible value. considerable time neces- CIRCUIT-BREAKERS 103 sary to insert the resistances and to dissipate a large amount of energy therein. (b) By greatly extending the are and using some efficient method of cooling, say by air blast or forced liquid, to obtain a large voltage gradient in the arc. But even with a voltage gradient of say 200 V/cm, which is high, the required length to break a 100 kV are would be five metres. (c) By bringing the current to zero artificially by superimposing a current oscillation on the d.c. Some proposals have been advanced to achieve this 3955) and several experiments conducted. Though this is technically feasible, it is prohibitive in view of the necessary size of equipment. To explain the principle, one method is described here. Valve V, is connected in series to carry the normal current J,, Figure 6.11. Valve V2 is blocked by its grid. Capacitor C, in series with Vs, is charged to the opposite polarity to a sufficiently high voltage (its full voltage rating), by some auxiliary charging equipment. When short-circuit takes place, the valve V2 is opened by its grid and the capacitor is con- nected to the line. The opening of valve V2 is followed by a process of commutation. Since the capacitor has previously been charged negatively, a negative voltage would be produced across the fault which would reduce the current in the valve V; to zero. The tendency at this stage is for the current to reverse but it is maintained at zero by valve action the voltage across the valve V; being negative. Current flows into C, which charges until its polarity passes through zero and becomes positive; this will, in its turn, produce a positive voltage across V;, but in the intervening period its grid will have been given a negative bias and will thus have regained control. Vj thus ceases to conduct and isolation of the faulty line is accomplished. C still continues to charge until full d.c. voltage is established across it; thereafter it must be isolated, discharged and recharged again to the opposite polarity before the system can again be operative. This is clearly a serious practical disadvantage. Other difficulties inherent in this type of approach may be seen by considering the size of capacitor which would be required. Its size would be determined by its need to store energy corresponding to that in the d.c. reactor and transformer inductance whilst carrying short-circuit current, and its ability to produce the negative voltage across V, for a time sufficiently long for this valve to deionise and regain control of the circuit. Gosland“3) has suggested that for a d.c. circuit inductance of similar value to the transformer reactance and a short-circuit current of four times the full-load current, the capacitor for switching purposes on a 200 MW transmission scheme would occupy a space approximately 10/3! It may be thought that the short-circuit current would not rise to four times the full-load current. But in a tie line which has small capacity compared to main line, the short-circuit current in the tie line could rise to even higher'values. The cost of such a type of circuit- breaker has been calculated by Busemann“#) for different ratios of short-circuit current/ normal current and different percentage reactances, and is undoubtedly prohibitive. The valves V; and V2 would have to be designed to withstand voltage stresses higher than the { line voltage and Vz would in addition have to be designed to carry current pulses of large magnitudes. Such a circuit-breaker is clearly impracticable. Russian engincers are, however, continuing with their research in the field of d.c. circuit- breakers, and new solutions may emerge.@10) | eer Mi co | * oe] an treme, erro ia nahn item seat Fee 2 cee 104 “3 ne Figure 6.11 Capacitor type d.c. circuit-breaker 6.11 Possible solutions to the problems of switching circuit-breakers(t12 Although circuit-breakers are not available, there appears to be no good reason why interruption by means of rectifier grids should not be suitable within acceptable time limits for stability. Since the d.c. system will invariably be of large power, the maximum limit of this time may be taken to be approximately 300 milliseconds. The following are possible methods of approach: (a) If the rectifier angle £ is increased to more than 90° (less than 180°) then the rectifier becomes an invertor with the current flowing in the same direction as before. On occurrence of fault, a rapid phase advance can be given to the rectifier so that it absorbs the energy stored in the reactor instead of supplying power to the faulty line. Conventional air-blast circuit-breakers (provided at both ends of each line) then open shortly after the phase advance has taken place to isolate the faulty section. Immediately after this, the angle can be reduced to that for normal rectifier working and power transmission rapidly restored. [t is implicit in such a scheme that the circuit-breakers are capable of current chopping at some known and predetermined level which is clearly related in time to the system parameters. (b) A further method is the blocking of as soon as possible after fault. bridge which will then stop cond without the use of high-capacity d.c. 4 \ a rectifier and the opening of its by-pass valve The by-pass valve will take over current from the ucting. The by-pass valve will continue conducting through the short-circuited line until the energy in the reactor is dissipated; the time constant for dissipation may well be large since the L is mainly the inductance of the reactor, and R may be very small since it is the resistance of the path through which the fault current is circulating. The insertion of resistance in series with the by-p; creasing this time constant, but would at the same ti to by-pass valve more difficult. Suppose tl 6.2 (c)] when valves 3 and 2 were conducting, ass valve would have the effect of de- me make the commutation from bridge hat the blocking took place at instant 4 [Figure . After point C, the by-pass valve would short- Te | = a a tek ay e 2 G 5 MMe. the POSSIBLE SOLUTIONS OF THE D.C. SWITCHING PROBLEM 105 circuit phases B and Y and commutation would start. It is quite clear that this commuta- tion should be finished before point D, otherwise valves 3 and 2 would fire again. Taking into account the deionisation time, and allowing a margin of safety, the angle of commutation should not be more than, say, 90°; since the resistance will increase the com- mutation angle, there will be a maximum limit to the resistance which can be inserted in series with the by-pass valve. A major objection to the insertion of this resistance arises when bridge units are con- nected in series. When one bridge of such an arrangement is blocked, the by-pass valve of that bridge has to carry the fuil current of the other bridges and it is undesirable to insert resistance directly. A way round the difliculty is to provide a short-circuiting switch across each resistance as shown in Figure 6.12. These switches remain closed normally and also when one bridge only has to be blocked. When it comes to blocking all the bridges due to a line fault, the switches should be opened immediately before the blocking action, thus inserting resistance in series with the by-pass valves. Alternatively these switches can normally remain open, and also when all the bridges have to be blocked due to a line fault. When it is a matter of blocking one bridge, only the corresponding switch can be closed. (The advantage of this method over the previous one is a saving of time which would occur in opening the switches on the occurrence ofa line fault.) The switches open at zero current, and although they must be fast acting, they perform only an isolating function. As mentioned above, the faulty line is subsequently discon- nected. The size of resistance to be inserted in series with the by-pass valve can be calculated as follows. The commutation of current from bridge valves to the by-pass valve starts with the short- circuit of two phases by the by-pass valve and finishes when the short-circuit current has attained the value J, (assuming that the reactor maintains the line current to Jj). Com- mutation starts from voltage zero. The equation for the short-circuit is: 2 +R V2E .SiN Wh, voce cecccceceeceeseeees (6.1) Bridge Bridge circuit circuit Figure 6.12 Mcthod of inserting resistance in serics with the by-pass valve 2 7 etn ann hla tt ks A 106 PROTECTION OF H.V.D.C. SYSTEMS — where R is the resistance of the short-circuit and is mainly the resistance in series with the by-pass valve. The solution for iis: VE ; - 1 Rei del -[R sin wt—2wL.cos wt]+C.e Whemt 10, IO, o0 000.1 e ec eceec esse eeneeecesnstvenesesseseeneeccecees (6.2) 20L.V2E and C= Re aakL Tete e eee een cee cee steer ncesensecectsstatsececucs (6.3) 2E ; -Rt wing : [ asin wt—2aL.cos wt+-2wL.¢ 2 | seedeteeedetebee pete ele eee (6.4) Also, when wt=y, i=I,. 2E : -2 2 = on [R sin y—2wL .cos y+2wL. ¢ 2° | sle-araisltaveleleds wlan le Chalgds & (6.5) It may be noted that by neglecting the resistance the above equation reduces to: "a= ToL (1—cos y), which is the same as equation 3 of Chapter 3. Assuming an allowable value of y and from the known values of /,, E and wl, Rcan be calculated from the above equation. In fact the maximum allowable value of R will be considerably greater than wL and hence can be easily calculated approximately by neglect- ing wL terms. Assuming that y=90°, equation (6.5) simplifies to: I = dries re eilels aaa eles ht care cdtled tte ere eI IELALE UA PEEL LL | (6.6) V2E_aV, Ps ed des yeteer ale AHMET aa aU UU (6.7) For ¥,=100 kV and [,=100 A R_=1000 ohms approx. In fact considerably smaller resistance will be sufficient for the above purpose. To adopt these methods it will be necessary to detect a line fault as quickly as possible. This can be done by utilising an integrating circuit in the protection arrangements and so measuring the rate of fall of voltage on the d.c. side beyond the reactor. As already considered in Section 6.6, it will be necessary to avoid invertor faults being mistaken for line faults by setting a discrimination limit for invertor faults. Since we are concerned with several circuits, it will also be necessary to detect the circuit which is faulty by measuring the rate of rise of current in each circuit. The total current will not rise substantially, but a greater proportion than normal will suddenly appear in the faulty circuit. As considered before, air-blast citcuit-breakers will have to be arranged to trip subsequent to blocking. Automatic reclosing arrangements could be provided to allow for the majority of atmospheric line faults being of a temporary nature. 6.12 Overvoliages Overvoltages may conveniently be classified as internal or external. eens EE I A A ER A ee ~ - . or ab OVERVOLTAGES 107 6.12.1 Internal overvoltages These overvoltages are the result of various operating conditions of the system. They may be caused by invertor and rectifier faults, blocking and unblocking operations, starting and interrupting the d.c. transmission, etc., and are usually the result of the oscillatory effect of the capacitances and inductances of the system. Depending upon the damping, the duration of these oscillations may be a few cycles. The frequency and amplitude of these oscillations, and the amount of damping for different types of interruption and for various operations with the same type of interruption, depends upon the action of the protective and control devices. There is thus an important ficld of investigation to find out how, by means of suitable protective and control: gear, these oscillations can be damped to a minimum. According to Lamm,99) when starting up the Swedish transmission scheme the rectifiers and invertors could be started with grids blocked, and unblocking the grids resulted in a smooth rise of voltage giving no overvoltages during the starting-up period. It is essential that the insulation of the line and other equipment should be able to stand internal overvoltages (as in the case of a.c.). According to the experience of the Soviet Union,“° by providing automatic grid control operation during transients together with suitable damping circuits, the line voltage does not exceed 130-150 per cent of the working voltage under conditions of normal switching operations and disturbances; even in the cases of some of the rarest combinations of faults, the largest overvoltages do not exceed twice the normal working voltage. In the design of the Stalingrad-Donbass line,“ surge diverters are installed at the poles of the line as a protection against overvoltages and serve to limit the maximum overvoltage between the poles and earth to 170 per cent of the line voltage. The insulation of the line and substation equipment connccted to the line is chosen for such a voltage level. Internal overvoltages caused by the commutation processes inherent in converter opera- tion, are discussed in Section 6.13 6.12.2 External overvoltages These overvoltages are mainly caused by atmospheric conditions and may be in the form of direct or side lightning strokes, statically induced overvoltages or similar other reasons as in a.c. practice. To protect the overhead line against lightning overvoltages a continuous earth wire on the top of the towers should be provided. This not only gives a shielding effect but also reduces the wavefront steepness and the amplitude of the surges on the line due to electric and magnetic coupling. It is also important to provide low tower-footing resistance in order to reduce the severity of lightning and avoid back-flash. Continuous counterpoise (buried conductor connected to the tower footings) can also be provided, where necessary, but this is expensive. It may also be desirable to provide rings and arcing horns across the insulators. It must be remembered that any flash-over across the insulator or horn-gap has to be followed by blocking the rectifier grids, since a d.c. are, once struck, will not be extinguished until the current is interrupted by the rectificr. Lightning’ arresters can be provided (and have been developed in Sweden and Russia) 3 4 pn nin nianninianiinnintaatsintabamnisaciinrss ss Lik 108 PROTECTION OF H.V.D.C. SYSTEMS very close to the terminal stations. With d.c., capacitors can be provided as surge absorbers at a greater advantage than in a.c. practice, since a fairly large capacitor can be provided as an cconomical surge diverter with an additional advantage of smoothing and reducing harmonics. The capacitor can be installed immediately after the line inductance and can be protected by a spark gap against the rare occurrences which cause blocking of the rectifier, This arrangement with a quick acting grid control has been found adequate in Sweden and the d.c. lightning arresters are not considered necessary.“109) For the Stalingrad-Donbass scheme as well, lightning arresters have not been considered necessary.(109) If there is a short overhead line from the converter station to the earth electrode, it will be necessary to provide some protection against lightning. A horn-gap type, d.c. lightning arrester developed for low voltages 6) can be used. ; At high operating voltages, for a line with an overhead earth wire and low tower footing resistance, the side as well as the direct lightning strokes become rare and of secondary importance. Thus in h.v.d.c. transmission lines, the choice of insulation level will be entirely based on the ability of the insulators to withstand operating voltages and switching over- voltages. 6.13 Parasitic oscillations 6.13.1 General During commutation, sudden changes take place in the voltages across various parts of the converter. The greater the firing delay, greater is the sudden voltage change. These sudden voltage changes cause oscillations in the transformer windings and other auxiliary equipment. These oscillations produce three effects: (a) Extra voltage stresses on the valves causing— (i) Backfire in rectifiers. (ii) Grid blocking failure which results in commutation failure. (6) Current surges in the valves. i (c) Disturbances if these oscillations fall within the radio and audio range. There are two stages of commutation: (a) When commutation starts. (6) When commutation ends. . At both these stages a sudden voltage change takes place. 6.13.2 Voltage stresses (Figure 6.13) Consider rectifier operation, and when valves 1 and 2 are conducting together. (@) When valve 3 fires, commutation starts and two changes take place: (i) Voltage across phase R increases by half the commutating voltage. This increase is caused by the commencement of the fall of current and thus induces a positive voltage. (ii) Voltage across phase Y suddenly falls by half the commutating voltage, caused by the commencement of the rise in current. Obviously sudden changes cause oscillations in phase inductances, which have capaci- tances across them, between phases and with respect to earth. Low resistances of windings provide very little damping to these oscillations. fi t [ ssorbers tgrided cbicing ectifier. iF and bass on 3 of hese Nary | { Ig a 2 4G re to ¢ a PARASITIC OSCILLATIONS 109 Earth capacitances and inductances of valve auxiliaries Ly \ 5 WOO Earth capacitance of the transformer windings (a) Figure 6.13 (a) (6) Parasitic oscillations during the commutation process Apart from this, capacitances and inductances’ of various auxiliaries cause oscillations which may well-be at a number of frequencies. These oscillations will appear across the valves which are not conducting, superimposed on their back voltages, and will cause severe stresses on these valves. On valve 4 these oscillations will appear at a, as shown in Figure 6.13 (6). On valve 5 they appear at b and on valve 6 they appear atc. Atc the voltage is reduced, hence oscillations are not very dangerous. Ata and b these oscillations, if large enough, could cause a backfire. : (6) When commutation stops, i.c., when valve 1 has stopped conducting, the voltage in phase’ falls by half the commutation voltage and in phase Y it rises by half. It must be noted that commutation voltage is considerably larger than when commutation Started. In this case valve 1 will be in the worst condition (i.e., it is the valve which stops J a ann en liens SANS Seah SCTE oe eS eyvamaaaicdle 110 PROTECTION OF H.V.D.C. SYSTEMS fo conducting), in which the back voltage and oscillations appear at d. Since it has stopped conducting, the oscillations of phases R and Y will add up across it and will result in oscilla- tions of increased amplitude. Apart from this, valve 1 has just stopped conducting and is fully ionised and hence these stresses on the anode will be very severe. Oscillations on valves 4, 5 and 6 will appear ate, f, and g respectively. Oscillations at e and fare also quite serious, since the back voltage is near its peak. In the case of invertors, these oscillations will appear across the valves superimposed on the positive voltage, and may cause fire-through. These oscillations can also appear in the grid circuit and introduce parasitic pulses on the grid and may cause a valve to fire too carly. Apart from backfire, such oscillations can also cause flashover on the anode in- sulator which will have an effect akin to that of an external backfire. When there are two or more units in series on the same side of earth, the frequency of oscillations will be different in the higher potential unit(s) than in the lower potential unit (earth side unit). This is mainly because, in the lower potential unit, two (when commuta- tion starts) and one (when commutation ends) of the three capacitances to earth of the transformer phases are short-circuited by the conducting earth-side valve(s): Thus the equivalent oscillation circuit for the higher potential unit(s) will be as shown by Figure 6.14 (a) and its approximate oscillation frequency will be I fue 5 Jz Sata fee cee eecttteeeees (6.8) where L is the transformer inductance/phase and C is the capacitance to earth/phase. The equivalent circuit for the lower potential unit, when commutation ends, will be as shown by Figure 6.14 (b) and its approximate oscillation frequency will be: FEY Ae LA | ek Pemx |S o=5° Five Piles dated teed (6.9) 6.13.3 Current surges Before valve 2 fires, the voltage across the capacitance of phase B (across valve 2) goes positive (Figure 6.13), When the valve fires, this positive voltage (say y;) across this capacitance is brought to zero. Energy stored in the capacitance discharges through valve 2. This situation may be severe; the capacitance forms an oscillatory circuit with the stray capacitances and inductances of the auxiliaries and a sharp current surge will be experienced by valve 2. Thus the higher the delay angle and the voltage, the higher will be the current surge, since the stored energy in the capacitor will be equal to cv,2/2. Thus these current surges will be much greater in high voltage valves than in low voltage valves, for the same current rating. These current surges may have a slope of well above 10°A/sec. Though this surge will be quite well damped, it could be large enough to cause quenching of the spot? (Chapter 12). Current surges can occur due to various other capacitances in the plant units. Such failures cause severe stresses on valves and cause commutation failures in invertors. Current surges may also produce high frequency radio interference near to substations. cr egeese qT: and is nyon qT: Lunt nula- fe en by [ (6.8) L Deas [ > Zz 2 = je i fo ria aa f = PARASITIC OSCILLATIONS ill L CG erie et L C | ascrllator TOW: —t-——H circuit of Mi C | top bridge i —|-—H unit L c + (8) fguivalent L c | oscillatory Tae —} AF circuit of a L bottom i —t———— earth side) bridge unit Figure 6.14 (a) (6) Equivalent circuits for voltage oscillations occurring at the end of commutation 6.13.4 Damping methods A number of investigations, theoretical and practical, have been made on these oscilla- tions and many methods of damping have been suggested. ) (0) (36) G7) Gs) Gos) The damping circuits for suppression of voltage oscillations consist of suitable values of capacitance and resistance in serics. These damping circuits may either be connected across each valve, that is six damping circuits, or in the form of four circuits, three for the transformer phase windings (in star or delta) and one across the d.c. terminals (Figure 6.15). A combination of both methods, involving ten damping circuits, may also be used. These damping circuits will not only serve to damp the oscillations but also to reduce the steepness of the sudden voltage changes. The following equations have been given by Busemann‘7) for the design of damping circuits consisting of resistance and capacitance in series. They correspond only to the voltage oscillations caused at the end of commutation, since these are much more severe than the ones at the beginning of commutation. If damping circuits are connected in delta across the transformer secondary, the value of each capacitance and resistance required for the higher potential unit(s), to obtain critical damping, will be given by: ne a a i el aR Me ey, 112 PROTECTION OF H.V.D.C. SYSTEMS 81 L . Bee Manel EIT 6 and R, J ce. (6.11) For the lower potential unit, CNBC | on oe ice tion vacleeebaee titel es dsleeeeesascslesatideble dc. (6.12) i 7 81 ZL and R'y= (Fa) SOOO reese ec ce en ee cee eeceereseceneesesessereseece (6.13) For star-connected damping circuits across the phase winding, the capacitance required A” will be three times the value obtained from either equations 6.10 or 6.12, above and the resistance will be one-third of the above value obtained from either equations 6.11 or-6.13. ! In practice, damping circuits with values smaller than those given by the equations above will be required to obtain satisfactory damping. dy a Figure 6.15 (a) (6) Damping circuits for oscillations Figure 6.16 shows the shape of the phase-to voltage with the sudden voltage changes superimposed, which occur during six commuta- tions per cycle, two of which reduce instantaneous the phase-to-phase voltage to zero at an angle a and the other four change, as shown, to half the instantaneous voltage at angle a. At each commutation the capacitor is charged and discharged and the resistance has to dissipate the energy stored by the charged capacitor. There are four changes per cycle with an amplitude approximately equal to 1/2. Esin a, and eight changes of half that value; thus the losses in the three damping circuits are given by: -phase voltage, consisting of sinusoidal ~ S hee Ra che hy " a 8 G fx 3 PARASITIC OSCILLATIONS 113 Figure 6.16 Phase-to-phase voltage of the transformer secondary when connected to a bridge unit 2 Py=33.4 Cp*=3.4Cy.f | 4 (V2. E sin a)2-48 (F .Esin ) |= S1Sf CC, E®. sin2 a 0.06. eke cocci ue caduecdnccscccsseeceee (6.14) If the damping circuits are connected in parallel to the valves, the critical damping is obtained for the higher potential unit, when, Ce eee Ce EIEN Ue Le (6.15) 45 7 and R,= (7-4) Heh tlle bi tele teste tokleh lel tdee sh dele LLL LI, (6.16) where C, and R, are the capacitance and resistance, respectively, of each damping element across the valves. For the lower potential unit C= nc paler departs ett ewleletdacuttaslewe tac thecsestes cle erdash i. deh (6.17) 1/243 R= EA PAP ffs ale wot diate wrassl be le le tole ale dalalele-dclie sleble b veelay sae olalecs dere . and R’, AG =) (6.18) The losses in the six damping circuits are given by: Pi=}f.C,.6. [2 (V2. Esin «)?+6(}. 1/2. Esin a)? |= =21f.C, EB? sin? ao... cece ee ec eee ees ede ddlet Meaaacketulellt tsk (6.19) Substituting the values of C, and V, from equations (6.10) and (6.15), respectively, into equations (6.14) and (6.19) shows that for the higher potential unit(s), the damping circuit in parallel with the transformer phases requires about the same losses as the damping cir- cuits connected across the valves. But substituting the values of C’, and C’, [given by equations (6.12) and (6.17)], instead of C,and C,, into equations (6.14) and (6.19) shows that for the lower potential unit the damping circuit in parallel across the valves have lower losses. Furthermore, if, in a bridge unit, valves are connected in series, these damping units can also be arranged to provide equal voltage division across the valves in series. Damping cquipment may be installed, as in the Gotland h.v.d.c. transmission, across the reactor in the d.c, circuit, and will serve also as a surge absorber. In other cases, e.g., on Bohne i Cid 114 PROTECTION OF H.V.D.C. SYSTEMS the English Channel scheme, the damping equipment is more suitably placed in shunt across the d.c. poles. In any event, such damping circuits are complex and expensive, and consist of combinations of R, L, C elements; the choice of series or shunt damping circuits depends upon the transmitted power, the level of d.c. and the transmission line parameters, Further to this, horn-gaps can also be provided across each valve, to protect against local overvoltages. This arrangement, if it flashes over amounts to an external backfire in the rectifier and a fire-through in the invertor; it is, however, a better alternative than the risk of valve damage. I To limit the current surges, it has been suggested (0) that low capacitance high frequency chokes should be connected in series with each bridge arm, as shown in Figure 6.15 (6). The slope of the current surge should be damped to below 106A/sec (Chapter 12). In the Stalingrad to Donbass h.v.d.c. project,{°) an air-cored reactor of 1.5 mH inductance has been provided in scries with each valve of 50 kV, 900 A rating. aa inn aha it iain inh dc neti taut hid eset i alas aaa east of CHAPTER.7 vad t : cuits Pe Reactive Power Requirements t e : References (49) (50) (61) (105). L 7.1 General risk As discussed in Chapter 3, invertors operate invariably with a leading power factor, the 4 value being given approximately by: #7 ie (cos 8-+cos f) cy a 2.22 (). | 2 (2.22) Commutation must be complete by an angle 5, before the voltage changes sign, and should be so under conditions of fault and voltage fluctuation. The requirements for reactive power are very dependent on the regulation which is provided and, with this in view, it is convenient to analyse uncompounded invertors first and then to take the case of com- pounded invertors (see Chapter 5). a uw 7.2 Uncompounded invertor It is clear that an increase in commutation angle accompanies a decrease in voltage on the a.c. side and an increase in direct current. Changes such as this are likely when there is a fault on the a.c. side, since this will not only reduce the voltage on the a.c. side in the normal way but will increase the direct current because of reduction in the invertor back-vollage. It has already been explained that the angle of firing f will increase automatically for such a condition in a compounded invertor; this cannot beso in the uncompounded invertor. It is necessary in this case to provide a large angle 8 during normal operation so that no invertor failure takes place for a considerable reduction in voltage on the a.c. side. Continuity of supply to the a.c. system is thus assured and is particularly important from the point of view of a.c. system stability under fault conditions. By providing a large angle B during normal operation some safe limit of voltage reduction can be provided so that, up to that limit, the commutation is still completed by an angle 8, before the change of voltage sign.49) : E I= (Cos SB—COS PB)... ccc eee e cence eee eee bee eens (3.18) V20L, cos 8=cos B+J,. viens Oye he eee ee EU UEELL I (3.22) ' 2 -*. COS pal eel tr, ies er 008 BT y, ae eee eeeceeenaceres Melee gale sleet docta loth ol ctoleh edlele (7.1) Again from equation (3.18), V/2wLe " cos B=cos §—I, 116 REACTIVE POWER REQUIREMENTS cos § cos § ols *. cos f= 2 +3 4/2E =cos §—J, = sles dete oa coreltct o alelepeileret h diols ised dala d dteleaditaleh t (7.2) Now suppose that a safe margin is required during the period of reduction in voltage on the a.c. side from E to xE and increase in d.c. from [,to yI,. Then the angle £, during normal operation, should be such that when the a.c. voltage is xE and the d.c. is yly, should be at Icast equal to 8,. A Ty. ; .. cos B=cos 3, V2Y la ole ogee beg eae fecaat erect les eeseie'e ace clesect cic) (7.3) : a The power factor during normal operation is given by: : wLe ; cos ¢=cos B+J,. OE TTT ce neces cence (7.1) i By substituting cos 8 from equation (7.3) into equation (7.1): I, . wl, (@ ) . Se a ees eet eet eee sect 0 ce etes ee cbc 7.4 V2E : (7.4) x If 5V/V, is the ratio expressing the inductive voltage regulation due to the commutation process of the invertor, then 5 V_ /3aL, m7 \ ols : (= ; 2) . (Se) = SS, from (3.1) and 3.6)... ceeeees (7.5) ° Tv Substituting from equation (7.5) into equation (7.4): 8 2; cos $=cos ae a srefelg tee a jelerdce sce teci-tietetetece cafe atareiee eis tate tale (7.6) cos $=cos 5,— By way of example, if 8,=10°, 770.05; y=1.5 and x=0.8, then 3 =0.985—0.05 ( —— cos $=0.985 005 (5 ') =0.835 tan $=0.6 6 sin $ _Reactive power cos# Active power and sin $=0.55 Hence the reactive power which is required is 66 per cent of the active power in the a.c. circuit, and 55 per cent of the kVA of the invertor station. It should also be noted that the active power available does not exceed 83.5 per cent of the rating of the invertor. ‘ But this symmetrical reduction in three-phase voltages occurs only during three-phase faults on the a.c. side [Figure 7.1(a)]. During unbalanced faults on the a.c. side, however, phase displacement as well as voltage reduction takes place, although the voltage reductions will be lower than in the case of three-phase faults. Such displacement must be taken into account in order to maintain the necessary safe margin (it may be noted that single-phase faults on the a.c. side appear as two-phase faults on the valve side of the transformer). It can be seen from Figure 7.1 (6), i Er COMPOUNDED INVERTOR 117 AR 6Y 2B (@) Three phase Jault poe Ss (6) lwo phase fault ~ a ae / 3 Increased néduced angle _ angle eso ar 4 Figure 7,1 Changes in commutation angle and voltage due to a reduction in the invertor a.c. voltages that voltage reduction in phases R and B results in greater angle 8 for valves 3 and 6 and smaller angle for valves 1 and 4 (assuming that the instant of firing is unchanged), whilst causing voltage reduction for valves 2 and 5. Thus valves 1 and 4 are the ones most liable to fail. Thus it is clear that the reactive power requirements for an uncompounded invertor will be very high indeed. os x z 7.3. Compounded invertor The need fér proper compounding of the invertor for not only saving reactive power but a eee “ | 118 REACTIVE POWER REQUIREMENTS also for providing carr safety hus already been discussed. Ultimately the most satis- | factory solution is t¢ tain the exact firing angle for each valve to fire, as explained in [ Chapter 5. “ For a compounde< =vertor, 5 will remain constant, say 5,, and will vary as required. 4 For normal operatio= the invertor, af wl, a a 2 3 cos =cos §,—> JRE (7.2) a) ae c £008 $008 Bs eet cecvesv te titwitibiicedeseisiititii (7.7) al : For 8,=10° and —=0.05, as before, cos $=0.985—C.==0.935 and tan $=0.38, anc =r d5=0.36 Thus the reactive p>==required is 38 per cent of the active power, and 36 per cent of the KVA rating. The ac=-7ower available is 93.5 per cent of the kVA rating of the invertor station. During faults on the == side of the invertor there will be certain temporary overloads on the source of the reac—= s0wer; if x and y have the same significance and values as in Section 7.2, then cos $=cos 8,— cos £=0,985—0.0=.— =0.891 tan $=0.51, and t=<active power requirement is now 51 per cent of the active power under faux snditions. It is necessary to mak= rovision for the possibility of having to carry out current regula- tion from the invertor =: From the regulation curve (Figure 7.2), the working point may be shown by A. == ‘ower the point, the less will be the power factor and the more will be the reactive pow=<quired, though by means of a tap changer on the transformer operated by a power faz——velay, the power factor will not be allowed to go beyond some limit, say B (Figure 7.2:. bout 10 per cent extra reactive power may well be provided to meet this requirement. The reactive power re—rment depends largely upon the transformer leakage reactance, apart from the deioniss=m angle 5,. Hence from this point of view, the transformer leakage reactance shoviz -< small. This can best be achieved by connecting the syn- chronous condenser to = nw voltage (about 11 kV) tertiary winding [Figure 6.6(6)], such that the tertiary windins =s as small a leakage as possible compared with winding on the line side. The provisioz reactive power in this manner is very effective in maintaining voltage at the invertor <==¢ faults, enabling the invertor to supply short-circuit current and thereby improving == ability of the a.c. network,(60) ie me pth satin [ SOURCES OF REACTIVE POWER 119 Rectifier A Invertor a cS a6 L memes Figure 7.2 Reactive power requirement due to current regulation from the invertor side 7.4 Sources of reactive power The reactive power requirement of an invertor, with suitable compounding and with facility for regulation from the invertor end, may be about 0.5 to 0.7 kVAr per kW of active power supplied to the receiving a.c. system. This kVAr can be supplied either by a syn- - chronous condenser, a bank of static capacitors, or a combination of both. = The former has excellent characteristics from the point of view of regulation, but the a ; possibility of using static capacitors is worth considering since they have lower losses and j f are possibly cheaper. ; Generally, both types of equipment require a similar investment per kVAr but the losses ince, ' of the capacitors will be only of the order of 0.2 per cent to 0.25 per cent compared with 1.5 rnyety’ : per cent to 2 per cent for the synchronous machine. Furthermore, with static capacitors, ae there is also a possibility of a reduction in commutation angle, y; this may result in reduced _ reactive power requirements by the invertor. , the It is possible to run an invertor without any dynamic source of alternating voltage, “t: €.g., synchronous condenser or power system. In such a case, static capacitors are used, the system and invertor being run at the natural frequency determined by the capacitance provided and the parameters of the connected system. [ , . (rm 120 REACTIVE POWER REQUIREMENTS With such a system, a load change alters the frequency and voltage, and the difficulty can only be overcome by providing parallel connected saturable reactors. These reactors, however, imply an extra consumption of kVAr and such a method, although technically possible, becomes very uneconomical and complicated. If the h.v.d.c. link is feeding energy into a system which consists of one or more alter- nators, then static capacitors can be provided for the invertor. The disadvantage is that when the voltage drops because of a load increase, the current in the static capacitor and hence the kVAr supplied, also fall; this is accompanied by an increase in the reactive power requirements of the invertor. If a synchronous condenser is used instead of the static capacitor, then the regulation gear provided with it would increase the kVAr and maintain the system voltage. Thus if static capacitors are provided, the alternators of the system will have to carry out the voltage regulation. Furthermore, their capacity should be such that for a given voltage drop caused by the increase in load, they are capable of increasing their combined reactive power output toa greater extent than the sum total of decrease in reactive power supplied by the capacitor and the increase in reactive power required by the invertor. If this condition is satisfied, it will be advantageous to provide static capacitors for the whole of the reactive power required by the invertor as indicated by recent Soviet tests on the Kashira-Moscow system. Static capacitors have an additional advantage in such cases; under fault or sudden load- change conditions in the a.c. system, the static capacitors on account of their large time constant can sustain the voltage, in the period immediately subsequent to disturbance, better than a synchronous condenser, and will not allow sudden voltage changes. This action will enable the invertor valves to adjust their firing angle to correct values with greater facility and hence reduce the incidence of commutation failures. In view of this recent Soviet experience, the original proposals for supplying the reactive power of the Stalin- grad-Donbass system have been modified; instead of having a synchronous condenser/static capacitor combination, with ratings in the proportion 3:1, connected to the tertiary wind- ings of the main transformers, it has been decided to employ banks of static capacitors only, which can be connected directly to the 220 kV system bus-bars. This practice has made it possible to reduce the total demand for reactive power from 450 MVA to 400 MVA for a transmission capacity to 750 MW, and to increase the efliciency by 0.4 per cent.“105) The possibility remains of compensating for reactive power by means of artificial com- mutation; this will be discussed in Chapter 8, PA NOME Ma EN RE St Se aes a it animate nana aeetal en . | } ih, wer { 2 ipod ra atari iate man 2 panne neti ttt ban ce CHAPTER 8 Artificial Commutation References (22) (49) (50), (61) to (68) inclusive 8.1. General considerations The necessity for operating an invertor at a leading power factor, thereby consuming reactive power, implies the provision of a reactive power source on the a.c. side. Artificial commutation is the means by which commutation can be achieved even after the voltage zero on the a.c. side, thus not only making it possible for an invertor not to consume reactive power, but even to supply it to the system load. This applies also to the rectifier which consumes some reactive power and thus operates with a lagging current. Artificial commutation may be applied in this case also and it is then possible for commutation to take place before voltage zero. But, unlike an invertor, a rectificr has no inherent deionisation angle and can be run at nearly unity power factor without any risk of failure; it thus consumes little reactive power compared to an invertor. Although the idea of artificial commutation appears to be very attractive, various in- vestigations have shown that the expense involved in achieving it undoubtedly exceeds that of providing a reactive power source on the a.c. side. Many investigations since 1932 have achieved technical success, however, and the subject merits inclusion here. One method in which the invertor is provided with capacitors in series with its trans- former secondary windings (Section 8.5) is likely to prove marginally economical, if designed to run at unity p.f.; it is thus considered in more detail than other methods. 8.2 Forced commutation in two steps‘22) (3) (64) 7) Gs) The principle involved in this method is that commutation takes place to an auxiliary valve which is at a much more positive voltage than the valve of the bridge circuit which is next due to fire. The voltage of the auxiliary valve is so arranged that it changes sign and becomes negative, thus making it possible for the commutation to take place from this auxiliary valve to the next valve, even at a lagging power factor. The circuit diagram is shown in Figure 8.1 (a). In addition to the normal bridge circuit, this consists of two auxiliary valves 7 and 8, a capacitor C and a delta-connected tertiary winding. The wave-form diagram is shown in Figure 8.1 (6). Consider the instant when valves 2 and 3, in conjunction with phases B and Y, are concluding conduction. At point A, which is at a distance B’ from O, valve 8 fires and commutation from 2 to 8 takes place with current being conducted through valve 8, the capacitor, phase Y and valve 3. Commutation takes place because the cathode of valve 8 is maintained at a negative voltage by the capacitor C, which is charged to a voltage XD such that it is negative towards the junction of valve 7 and 8. Since the commutation voltage AD is very large, the time for commutation y, is very short. As conduction proceeds the polarity of the capacitor is changed; at point E, where the cathode voltage of valve 8 becomes more positive than the cathode voltage of valve 4, it becomes possible for commutation to take place from valve 8 to 4. This takes a time 122 () Voltage diagram Secondary current in phase R, H_— \\ (C) Current diagrams LJ | i | A ls LN 7m 1 Capacitor current, #3 b> Primary currene in phase R, ae fg = lot by Figure 8.1 (a) (6) (c) Invertor with forced commutation in two steps J J fc ot om < FORCED COMMUTATION IN TWO STEPS 123 v2 Which is considerably larger than y;. Valves 4 and 3 continue conducting through phases Rand Y and the polarity of the capacitor is maintained up to point F (voltage XF with respect to neutral), such that the junction of valve 7 and 8 (the anode of valve 7) is positive. At point G, valve 7 fires and commutation takes place from valve 3 to 7 in the same manner as from valve 2 to 8. Now current flows through valve 4, phase R, the capaci- tor C and valve 7 and the capacitor’s polarity is changed to become negative. At H, valve 5 fires and commutation takes place from valve 7 to 5, after which, at J, valve 8 takes over from valve 4, etc. The approximate power factor angle is 6=$’+¥/2, where ¢ is the angle during which the capacitor changes from negative to positive peak voltage; the time ¢ secs, corresponding to y, is given approximately by J,x t=2V,C, where V, [XD in Figure 8.1()] may | be about 20 per cent higher than the peak value of the a.c. voltage. The capacitor voltage rises considerably if the next valve, which is supposed to take over from the capacitor and auxiliary valve, fails to fire. Thus if valve 4 fails to fire, the capacitor will continue to charge until point G whereupon valve 7 will fire and cause a short-circuit on the d.c. side. The capacitor can’be protected by firing quickly the non-conducting auxiliary valve as soon as the capacitor voltage rises above a permissible level. The angle over which current is conducted through the transformer secondary windings and valves is reduced from 120° by the angle over which the capacitor C conducts through the neutral, Figure 8.1 (c). The current through the capacitor will be of a frequency three times that of the system. Since the current passes through the transformer neutral, it will cause a circulating cur- rent in the delta-connected tertiary winding which is of the same frequency as the capacitor current. If the tertiary winding has the same number of turns as the secondary winding, its current will be ip==i,/3. The current i, in the primary winding will be the sum of i, and i; and will have a considerable third harmonic content. The time 8, is the period during which the main valve must deionise, otherwise it will strike again; thus 8, has to have some minimum value. Also deionisation of the auxiliary valve should be complete in a time 8, otherwise short-circuiting of the d.c. side through the auxiliary valves will occur. For a given capacitor, the charging time will vary with the d.c. current passing through it; for a large current, the slope of the charging line increases and 8, decreases. Thus the minimum allowable value of 8, fixes the maximum value of direct current which is permis- sible. Again, for a small current the slope of the charging line decreases and 82 decreases; thus in this case the minimum allowable value of 8, fixes the minimum permissible value of direct current. It is quite clear that by increasing the value of the capacitance, the charging slope will decrease and the value of the maximum allowable current limit will increase, but also the value of minimum allowable current limit will increase, and vice versa. This difficulty may be overcome with more complex’ arrangements; the capacitor can be constructed in sections, in parallel; at low loads sections are disconnected and then recon- nected when the load is increased. This switching can be carried out during intervals when the capacitor is non-conducting, but involves still further complications. — 124 ARTIFICIAL COMMUTATION It has been calculated?) that the capacitor rating required for unity power factor will be of the same order of kVA as that required for natural commutation, although at a work- ing frequency of three times the system frequency. This reduces the value of capacitance required for the same kVA rating by 66 percent; but does not reduce the size and cost of the capacitor in the same proportion, mainly because of the increase in dielectric losses with frequency. Two extra valves of full rating have also to be provided. Clearly the voltage across all the valves will be higher than normal, since the capacitor voltage rises above the a.c. voltage. Apart from this, the valves will have to withstand sudden voltage increases of the order of AD, much higher than those which occur during natural commutation. Further, the auxiliary valves will have to withstand these sudden increases three times in every cycle, a factor which will reduce the valve voltage rating by a considerable amount. Other expense is unavoidable in increasing the efficiency of damping and smoothing circuits and the provision of special protection arrangements which are suited to forced commutation. 8.3. Forced commutation in one step(22) (67) Figure 8.2 (a) shows a circuit arrangement which consists of two bridges connected by two star-connected secondaries in phase opposition, and with their star points joined through a capacitor C. As shown by the voltage diagram in Figure 8.2 (6) the valve conduction sequence will be as follows: Consider that valves 4 and 3’ are conducting. After they have conducted for 60°, valves 4’ and 5 fire at the same time and commutation takes place from 4 and 3’, respec- tively. After a further 60°, commutation takes place from 4’ and 5 to 6 and 5’, respectively, and so on. The complete sequence of conduction is: 4 4’ 6 6’ 2 2 2 5 5 1 3 > Each valve conducts for 60°, and the current through the capacitor changes sign after every 60°, thus working at a frequency of three times the system frequency. The current and voltage across the capacitor are of the form shown in Figure 8.2 (c) (the commutation angle has been neglected). When valves 4 and 3’ are conducting, the current in C flows from N to N’. If N is taken as positive, the capacitor will be charged in a positive sense as shown. Considering the case of unity power factor, at point S commutation is supposed to take place from 4 to 4’ by short-circuiting phases R and B’ (also from 3’ to 5 by short-circuiting phases R’ and B). Since the capacitor is charged in a positive direction, the voltage between phases ‘R and B’ will be obtained by adding the capacitor voltage to the voltage of phase R. Thus at point S, the commutation voltage will be the capacitor voltage under zero power factor conditions, and will be the capacitor voltage minus the voltage between phases R and B’ under lagging power factor conditions. Commutation of the other valves takes place ina similar manner. Since the deionisation of, say, valve 4 should be complete before the voltage between . phases R and B’ equals the capacitor voltage, there will be a minimum limit to the current SOLS ete eet AA a rCipr will t a vork- paeiuance d cost of i eeee tcross all | vettage. og cr of ther, the ‘e : 1B hing o forced ecg | by if ugh duction fof 60? TS] ectivery; al fler ictfrent- tutation ib is vi dering tg tto 4¥ ind tases KR eu rit tor auth B tee ina r= “oon Surrent (Th 2 FORCED COMMUTATION IN ONE STEP 125 1 0) oltage diagram 1 3R' 5B Sy" IR /B' 3y ) YF | @pacitor i V.| current 4 & voltage ny \/ \ Figure 8.2 (a) (6) (c) Invertor with forced commutation in one step [ iG — Spe. ws atehh 126 ARTIFICIAL COMMUTATION in order to obtain sufficient capacitor voltage. This minimum current limit can be increased by providing capacitance in sections and in parallel, as in the case of forced commutation in two steps (Section 8.2), and switching off sections at small loads. But since the capacitor is conducting all the time, this can only be done by providing a circuit-breaker. At small loads, an alternative to providing capacitance in sections is to supply leading current that consumes small amounts of reactive power. In this method there may well be a significant saving in capacitor kVA up to 40 per cent or 50 per cent, compared to the capacitor rating requirements for natural commutation. The output voltage V, can be obtained by calculating the average voltage over a 60° interval CD [Figure 8.2 (b)]. +7/6 3 E Vi=2.-] 1/2. = . g=2 ={v 2B cos wt. dwt —7/6 ~ where E is the voltage between phases, thus 2.76 eel lal : Vi= v' - E=1.56E, which is higher than the value of 1.35£ obtained from a single zs . bridge (equation. 3.1). Each valve and winding conducts for only 60° in every half cycle, thus the r.m.s. current J will be: +7/6 4 I= 7 . fra . dwt aT —7/6 : Ty, ¢ d to can arn single bridge are ss ina sin . “3B do COMP: Ji d g g The total are rating=2./3.E£.1 a = 1.28 Px Py Je 6 compared to 1.045 P,, ina thu bridge with natural commutation. Thus the transformer rating is increased by 23.5 per cent. The voltage across the valves will also increase because of the capacitor voitage. At instant £ [Figure 8.2 (b)], the voltage across valve 6’ will be PQ which is substantially higher than the maximum voltage across the valve in an ordinary bridge. The number of valves is doubled, but each valve conducts only for 60° compared to 120° in an ordinary bridge circuit; such a valve will not, however, be much smaller than in the normal case since the peak current is the same in both cases. 8.4 Resonant commutation®2) (62) In this.method a particular harmonic voltage at a suitable phase angle is superimposed on the fundamental voltage in such a way that it provides cxtra commutation area where commutation takes place and delays the voltage zero at which the commutation and de- Secotncneceanenenea a Crepe woe . = ~ ~ —~—- — . Se USE OF SERIES CAPACITORS IN THE SECONDARY WINDING 127 ionisation should be complete; this enables the invertor to work at nearly unity power factor. = : The current contains some harmonics, and hence by providing the resonant circuit in each phase of the secondary winding, as shown in Figure 8.3 (a), the voltage of the required harmonic can be obtained superimposed upon the fundamental. Figure 8.3 (6) shows the second harmonic superimposed on the fundamental; in this way it aids the valves only on one side of the bridge and hence is unsuitable although it is satisfactory for a half-wave converter circuit. This is clearly true for the cases for all even-order harmonics. Amongst the odd-order harmonics, the third is of no use, since it has no phase-to-phase voltage component. ~ Figure 8.3 (c) shows the fifth harmonic superimposed on the fundamental and it can be seen that it is quite appropriate for the purpose under discussion. A separate fifth- harmonic voltage for phase R only, is shown in the figure; seventh-harmonic will also be suitable. : From Figure 8.3 (c) it can be seen that the peak value of the voltage is increased because of the superimposed harmonic. Thus the voltage stress on the valves will be increased, thereby reducing their rating. The percentage harmonic content on the d.c. side will also increase, and for this a larger smoothing inductor will be required. Furthermore, the cost of the resonating circuits will be quite comparable to the cost of capacitors required for natural commutation, and automatic tuning arrangements for the resonant circuit will also have to be provided to allow for variations in the supply frequency. 8.5 The use of series capacitors in the transformer secondary windings®2) This method is the simplest and the best of the available methods of improving the power factor by artificial means and is marginally competitive with natural commutation. The circuit arrangement is as shown in Figure 8.4 (a). Figure 8.4 (b) shows the trans- former voltage and the d.c. output voltage, and Figure 8.4 (c) shows the voltage across the capacitors of each phase. Consider instant A when valves 4 and 5 are conducting through phascs Rand B; Cg is being charged in the positive direction, Cy in the negative direction and the capacitor Cy is maintained in a charged state at a negative voltage Ve. The commencement of commutation from valve 4 to 6 is possible as long as the cathode of valve 4 (that is the anode of valve 6, since valve 4 is conducting) is positive with respect to the cathode of valve 6. At point D, although the voltage e; (voltage between phases Rand Y) is negative, the voltage between cathodes of 4 and 6 is positive because of the supcrimposed positive voltage of the capacitors and is equal to veg+Vvey—e;, and will remain positive until ey=Veg+ Vey. . As can be scen from Figure 8.4 (c), the voltage veg+V¥cy remains approximately constant during the commutation period and starts to decrease as soon as the commutation is over; clearly the deionisation should be complete before e; becomes equal to vog+Vcey, say before point £ on the diagram. Since the voltage across the capacitor depends upon the magnitude of the current, the capacitor voltage at lower currents is liable to be insufficient. Thus the requirement is that the capacitor voltage at normal currents should be sufficiently high to give enough voltage at reduced currents for the same power factor; otherwise at co. GAR Zs GE ma Cm Sa ie aoe a IS VLR WS Sa TS, 128 ARTIFICIAL COMMUTATION (8) Circuit diagram fundamental + 2nd harmonic Fundamental, (®)Superimposed second harmonic Fundamental +5 th harmonic R goneoment (C) Superimposed fifth harmonic Figure 8.3 (a) (b) (c) Invertor with resonant commutation one | 4 { " | panies a a USE OF SERIES CAPACITORS IN THE SECONDARY WINDING 129 lower currents the power factor angle ¢’ will have to be reduced, and a correspondingly reduced percentage supply of reactive power will ensue. The capacitor voltage rating can again be kept low by providing the capacitor in parallel- connccted sections and switching off the sections as the load decreases. A safeguard against momentary changes of current can be provided by suitable compounding, which adjusts the angle of firing 8’ according to the current and commutation voltage available. With such an arrangement it is sufficient to design the capacitor so that there is enough capacitor voltage at 0.9 to 0.8 of the maximum rated current, for the same power factor. At the end of deionisation (point £): ey =V2E. (e424 3.) vor=Ve 2V¢ (8+ 7/2) voy Ve 720. where angles ¢’, y and 8, are in degrees. For successful completion of commutation and deionisation, at point E: Yor tVey>e1 ot 112), OR sin ($'y1243,) vececcccccseececcceceeerceceeees (8.1) We-2V 10 18000 .L. J, Furthermore, 3 ¥ Gin degrees) = ———_—4 ‘com where ¢,,,,, is the mean commutation voltage com wl V V-2V, Ccom [v+( 2) #)- V2.Esin ¢’ =2),. (1-28)- /2.Esin 120 3 18000. L. Jy es > te -?)- V2. Esin $’ 120 ave! Cc where C is the capacitance in each phase and ¢ is‘the time in seconds which C would take for charging upto a voltage of 2V, =1/150 sec. oes 8.3 “Tog ce (8.3) r.m.s. value of the capacitor current 2 I.=I= ALi sesh dle (8.4) The total capacitor a.c. rating Lt V. . ‘ ‘ , Ped. L=V3.Ve-Ta-- 85) (@) Circuil diagram Figure 8.4 (a) (See next page) — = eee Eee Ke i a i i RU ucla deaths dn Ge we then 130 /® Pansformer and output voltage (c) Voltages across oulpul capacitors Yolkage due to ~ lransformen Total valve voltage \. @ Voltage across \ valve 4 (commutation \, angle neglected) \ - . 2M | al, cL Voltage due to capacitors Figure 8.4 (a) (6) (c) (2) Invertor with series capacitors in the transformer secondary | = USE OF SERIES CAPACITORS IN THE SECONDARY WINDING 131 SS — From the above equations, the capacitor rating can be calculated. Consider an invertor arrangement in which BV E=100 KV, 1,=200 A, 7 =0.05 and 5,=10° o WV oL.ly Vi, V2E «.L=112 mH. Considering the case of unity power factor (¢’=0) and substituting the values above in equation (8.2): 2x 10-8 i = BROOD 26 112 NO 0 eet abtiteiatai (8.6) [equations (3.3) and (3.4)|=0.05 3 [ 3 2 ow, (1-24) 120, [ Assuming V,=30 kV, _ ke Y 2 —7° 2 The right-hand side of the inequality (8.1)=/2 . 100. sin 17°=41.5 kV 17 The left-hand side of the inequality (8.1)=60 . (1S) =s15 kV .. the condition expressed by (8.1) is satisfied. At I’,=0.97, V’=300.9=27 kV and from equation (8.6), Via 5a7.1 The right-hand side of the inequality (8.1)=41.7 kV The left-hand side of the inequality (8.1)=46.3 kV -. (8.1) is again satisfied. At I’,=0.8 Jy V’,=30 .8=24 kV and from equation (8.6), = Y Y=7.2° 2 The right-hand side of the inequality (8.1) =41.9 kV The left-hand side of the inequality (8.1)=41 kV «. (8.1) is not satisfied for the case of 0.8 Jj. Thus commutation failure will occur at 0.8 J, if the power factor is maintained constant at unity. sae fina From cquation (8.5): P.=/3 . 30. 20010400 kVA ER == fr & i eee 132 ARTIFICIAL COMMUTATION Active power=1/3.E. it ti ee [from equations (8.4) and (8.5)]= active power a - 100=V/1.5. a . 100=36.75 per cent. With natural commutation: -. ratio, Vv : Cos =Cos 5-5 (equation 7.7) o =(Cos 10°)—0.05=0.935 reactive power active power Thus it can be seen that the percentage capacitor rating is almost equal to the percentage saving in reactive power. At lagging power factors the capacitor rating may be somewhat higher than the reactive power saving. The advantage of this method, when designed to operate at unity power factor, is the increase in active power rating. It is quite clear that the active power rating of the invertor with natural commutation will be 1/3. E. I. cos ¢ and will be 93.5 per cent of the kVA rating (in the case considered above). The invertor with scries capacitors, and designed for unity power factor, will have an active power rating of 1/3. E. J, ie., 100 per cent of the KVA rating, which is an increase of 7.62 per cent. Thus the utilisation of the invertor station is increased. Figure 8.4 (d) shows the voltage across valve 4, which is the sum of the voltages due to the transformer and the capacitor. It can be scen that there is no increase in peak value and thus there is no increase in the voltage stresses across the valves. The output voltage diagram is shown by the dotted line in Figure 8.4 (b); this is also the sum of the voltages due to the transformer and the capacitor although there is no change in cither the mean value or the waveshape of the output voltage. Thus the voltage har- monics on the d.c. side are similar to those arising in the case of natural commutation. This method does, however, introduce a number of other problems. If there is com- mutation failure, the voltage across the capacitor may rise above its rated voltage. Pro- tection against this can be provided by opening the by-pass valve and short-circuiting the direct current side. It is unlikely that the invertor would be able to resume its normal Operation automatically, as it could in the case of natural commutation if the firing of its next valve took place correctly after a commutation failure (Chapter 6). Starting may also present some difficulties. These problems need further study and if operation under fault conditions is reasonable cnough, this method of artificial commutation is well worthy of consideration. tan ¢= =0.38=38 per cent. ed 5 > ae CHAPTER 9 Use of Earth and/or Sea Return References: (7) (12) (53) (69) to (74) incl., (113) to (117) incl. 9.1 General From the point of view of economy of transmission the earth, or sea, or both earth and sca are attractive media for the flow of return current. There is saving not only in the capital cost of the transmission line or cable but in the copper losses since the voltage drop in the return carth path will generally be considerably less than in the case of a metallic conductor. By way of example, the resistance of the return path in the Gotland scheme is 1 ohm only, compared to a resistance of 19 ohm approximately in the d.c. cable. The saving in conductor cost is offset by the need to provide the positive (anode) and negative (cathode) electrodes, and low voltage lines from converter stations to the electrodes, in the return (earthed) circuit. With a d.c. system using two conductors with earthed mid- point, the earth can be used as a temporary path for current in the event of one outer con- ductor being faulty, thereby increasing the reliability of the system. There are, however, a number of technical problems attendant on using an earth return path. These are: (i) The possibility of interference with low-voltage, d.c. railway signalling relays. (ii) Metallic corrosion of equipment in contact with earth, e.g., cable sheaths and pipes. (iii) In the case of sea crossing, the possibility of magnetic compass error in the neigh- bourhood of a submarine cable carrying d.c. (iv) The likelihood of harmonics from the d.c. transmission system causing increased inductive and capacitive interference in communication systems. The difficulties in (i) and (ii) may be avoided by finding a suitable position for the elec- trodes; this position may not always be near enough to the d.c. transmission system to justify the economy of using an earth return. (i) (ii) and (iii) are discussed below in this chapter, whilst (iv) is discussed in Chapter 10. 9.2 Electrodes and their arrangement The conditions for the two electrodes are markedly different since it is only the positive one (anode) which is susceptible to substantial loss of material through electrolytic action. Furthermore, oxygen and chlorine, which are highly corrosive, are released at the anode and this electrode should be made accordingly out of non-corroding material. On the other hand, hydrogen is released at the cathode and this gas tends to have a protective influence on steel, copper and other metallic materials. The resistance of the electrodes to earth is very important and should clearly be as low as om —_ 134 USE OF EARTH AND/OR SEA RETURN possible in order to minimise losses and reduce electrode heating; a high operating tempera- ture for the electrode accelerates corrosion and disintegration through thermal agitation, The resistance to earth can be decreased by increasing the number and size of the electrode elements and by ensuring proper contact with earth. According to Gosland,©3) the tem- perature of the electrode should not rise above 100° C provided that the moisture content of the soil is near to saturation. In a soil partially dried out, earth electrodes must be operated at a much lower tempera- ture to avoid movement of moisture away from the heated electrode because of the reduc- tion of capillary forces with increase of temperature. In this connection it is necessary to take account of short-circuit current and the time for which a short-circuit prevails; due to the fast-acting constant-current regulation of the rectifier (discussed in Chapter 5), the short-circuit current will not exceed 1} times the full-load current and will not last for more than I sec. For an electrode to carry 500 amp, it would be necessary to attain a figure of less than 0.1 ohm for the resistance to ground; this requirement is clearly necessary for both electrodes. The phenomenon of electro-osmosis has an important influence on the arrangement of the positive electrode.6) Electric current tends to carry water from a positive electrode to a negative electrode; with electrodes far apart, the result of this action will be the driving away of water in the vicinity of the positive electrode and the reduction of the water level in the ground. Fora current of 1,000 amp, the transportation of water may be of the order of 0.5 to 15 m3/hour. If the positive electrode is situated in ground with no constant supply of water, the surrounding volume will dry up and increased resistance and high temperature will ensue. A suitable arrangement in the earth is a clay hole filled with salt and positioned below the, water table to ensure a constant supply of water; the electrode must be made of non- corroding material for this case, as for an electrode immersed in sea water. If water is not available, or provided, then a very good metallic contact with the surrounding earth is necessary; the earth selected should be as good a conductor as possible. Electrodes situated in sea water have an effect on fish which tend to be attracted towards the anode and repelled from the cathode; some marine-life protection is therefore necessary. Experiments in Sweden showed 9 (0 that beyond 4 metres distance from the electrodes, with a current of 200 amp, there was no visible effect on fish; within 4 metres, fish were attracted towards the anode, and within 2 metres, fish died after about 4 minutes. Small fish were less affected than larger ones. These electrical effects are separate and distinct from poisoning of the water as a result of electro-chemical action at the electrodes. It is clear that eléctrode installations for sea crossings must include the provision of fences and breakwaters. Tests in Sweden, at Ludvika‘0 (0 and Trollhittan,©® (7) have been carried out to find the most suitable materials and surrounding media for electrodes. Cast iron, graphite, copper and magnetite (64 per cent Fe.03; 31 per cent FeO; 1 per cent 41.03; 4 per cent SiO.) electrodes have been studied in fresh water, salt water, clay soil, sand, marshy ground, and rocks. In all tests, negative electrodes were undamaged and showed no loss of material. Amongst the materials used for electrodes, magnetite showed the least loss since it is not attacked by oxygen. The following were found to be the most suitable arrangements: SO ne re Are EUR reer rt —~8oee 7 eee Spee eb Od a camer’ i = igs < 2 ELECTRODES 135 (a) Positive electrodes (i) Magnetite electrodes in salt water [Figure 9.1 (a)]. (ii) Tron rods in iron ore [Figure 9.1 (b)]. (6) Negative electrodes (i) Angle iron in clay soil or marshy ground [Figure 9.1 (c)]. (ii) Steel wire in salt water [Figure 9.1 (d)]. (iii) Iron rods in iron ore [Figure 9.1 (b)]. (iv) Magnetite electrodes in clay soil or marshy ground. Figure 9.2 shows the positive electrode arrangement provided at the Vastervik station of the Gotland link® (it should be noted that in this scheme, the polarity of the system is negative with respect to earth and thus the current in the sea flows from Vistervik, on the mainland, to Gotland). The electrode consists of twelve parallel-connected magnetite rods placed in an excavated basin on the sea-shore. Measures have been taken to prevent direct contact of the cable sheath and joint with sea water and a fence to exclude fish has been for fron ore 6 2 fronrod SN RE a @) ay all fF Positive or negative electrode iin iron ore \Magnetite Clay on morshy ground (2) Angle Negative electrode iron in the sea “+ a) Positive electrode in sea woter Fe (Cc) Negative electrode in clay or marshy ground Figure 9.1 Suitable types of positive and negative electrodes 136 USE OF EARTH AND/OR SEA RETURN Plastic cable Fence~, Protective tube Broken rock Cap of Neoprene Flectrodé. O0/2345M three toed Figure 9.2 Positive electrode arrangement at Viistervick (Gotland scheme) i& provided. The negative electrode of the Gotland scheme consists of a smooth copper i conductor laid on the sea-bed near to a rock reef at a distance of 350 m from the shore. Information relevant to high voltage d.c. transmission is available from the experience of Japanese engineers using the sea as a return conductor for schemes of d.c. railway electri- fication.419)_ This has been done to avoid corrosion of rails, some protection from electric shock, and to reduce the voltage drop in the return path; the electrodes are placed inthe sea and current up to 1,200 amp is diverted. Artificial graphite is considered to be a suitable material for immersion in sea water as an electrode on account of its substantial current- carrying capacity, insolubility, mechanical strength, ease of construction and economy. It has been experimentally confirmed in Japan (though not yet put into practice) that when graphite electrodes are given water-proofing treatment their lives are prolonged. Old, used rails placed in the sea water or driven into the sea-bed, and with lead connections attached above water level have been used successfully as negative electrodes. Tn a double-conductor, mid-point earth system, where earth is to be used in the case of failure of one side, the direction of current through the earth may be either way depending on the side which is shut down. Thus both electrodes must be constructed as positive electrodes. This is true also for a permanent earth system in which power direction is reversed by reversal of the current. If the use of earth return is short-term only, then the electrode problem is somewhat simplified and galvanised iron pipes are a possibility for the clectrode material. bE 9.3. Nature of current distribution in the earth When earth current flows from one electrode to another, it is likely to use pipes, cable sheaths, etc., thereby causing corrosion; it may also flow in rail tracks causing interference with railway protective relay circuits. In order to appreciate any particular case, however, it is important to know the main features of direct current distribution through the earth itself. , The earth is not homogencous and there is no question ofa uniform distribution ofcurrent. Swedish"experiments have been conducted to ascertain the nature of earth conductivity (17) and the results suggest a layer of the order of 1 km thickness with a resistivity of about 4,000 ohm-metres, a layer below this of primary rock with a higher resistivity of about 14,000 ohm- | | | pease a ~ eect cn neeneailananeanet nat tn RN EA A ELAS | i [ ; EFFECT ON RAILWAY TRACK SIGNALLING 137 [ Battery Shunt resistance |} Rheostat [. . ; J Relay Figure 9.3 ES Railway Choke—~ signalling arrangement J Re ‘ails N | \ | ve a metres, and then the molten mass with a resistivity as low as 800 ohm-metres. If the trl distance between electrodes is small, then current will be mainly confined to the uppermost te layer, the primary rock acting as an insulator. For medium distances, some proportion of i the current will pass through the inner molten mass of the earth; for large distances most ble of the current will take such paths. Resistivities, particularly of the upper layer of the Ne earth’s crust, are very variable from location to location on the earth’s surface. Compared f with the resistivity of the outer layer, that of a cable sheath may well be as low as 0.02 en ohm-metres. vd, : In gencral, it is clear that current density, and hence voltage gradient, decreases rapidly E : with distance from the electrodes. A voltage gradient in the surface layer which is suffi- 2 cient to cause large currents in nearby metal equipment will exist only in the proximity of of the electrodes; at long distances between the electrodes, the current will leave (or enter) the 7 electrode surfaces in all directions and this effect will thus be independent of the route. 7 The resistivity of the sea may be as low as 0.2 ohm-metres. Thus with electrodes placed Ge : 7 in the sea, very little current will enter the surrounding land; for long distances, with much Us land interspersed with the sea en route, the current tends to take a path from the sea into f the inner molten mass of the earth. 9.4 Effect of carth current on railway track signalling =] : +s A Figure 9.3 shows an arrangement of a low-voltage, d.c. operated relay circuit. When the 2 aod track is empty, the battery circulates current through the rails and the relay circuit; when 7 a the train arrives on the track, it short-circuits the rails and hence operates the relay which in its turn operates the signal or other necessary device. Itisclear that an unwanted voltage drop duc to earth current in the uninsulated rail may be sufficient to cause maloperations. ent In Sweden, where the Ieakage current in the track to cause faulty operation is about re 2.5 amp, an incorrect relay operation was observed as far as 45 km from the electrode for a current in the h.v.d.c. system of 170 amp. Calculations indicate that this distance will be 150 km from either electrode, for 1,000 amp, main current. Since this distance is large, it has been decided in the Swedish case to change the track relay circuit to one which is iar i | ! 1 Tier inn eine idea th nl . cata alli 138 USE OF EARTH AND/OR SEA RETURN insensitive to such d.c, interference.“ G) Low-voltage, d.c. operated track relaying is obsolescent; one system, called “Coded track circuit”,(73) which is also insensitive to stray d.c. interference, is satisfactory and in wide adoption. Difficulties of this type are greatly minimised if electrodes are placed in the sea. 9.5 Corrosion caused by earth current Earth return brings in very. serious problems of corrosion of sheaths of buried com- munication and power cables, that of its own cable, pipe lines, anchored ships and other metallic equipment in contact with the earth. Since the current in any of these cases will enter at one particular zone and Icave at another, the latter will form the anode and be the region at which corrosion takes place. Referring to Figure 9.4, it may be scen that different corrosion effects will arise with the two metal structures marked A—B and A’—B’ even though they are of similar dimensions and equally spaced from the positive and negative electrode, respectively. Thus the current will enter the structures at 4 and A’, and leave at Band B’ respectively: since B’ is nearer to the negative electrode than B is to the positive electrode, the greatest amount of corrosion will occur at B’. This applies particularly to cables and pipelines which are radially dis- posed in relation to the electrodes. : The leakage current per unit area is the significant quantity, and the amount which will be permitted is a matter of policy. A leakage of 1 A/cm? is about 30 coulombs per annum; and since the electro-chemical equivalent of lead is approximately 100, the quantity of this metal removed per cm?/annum would be about 0.03 gin.%3) This is small but there is always the possibility of concentration and the greater leakage current density accompanied by accelerated corrosion at some points. There is a difference in approach between the Soviet Union’s engincers and those of Sweden in calculating the leakage current density. Swedish practice has been to divide the measured or calculated leakage current per unit length by the cable surface area per unit Iength.“) The Soviet practice on the other hand has been to take half the cable surface area per unit length on the grounds that the entry and exit of the current will take place only from that half of the cable surface which is towards the dispersion field. The Swedish authoritics and those of the Soviet Union have decided on acceptable figures +ve eh de - iil vil i Metal structure Haat A Figure 9.4 Different corrosion effects in metal structures near to the positive and negative electrodes seiseaininaivta a s ca eat i nA nat Aha a a a a a a COMPASS ERROR 139 for their respective conditions. The figure of 0.1 A/cm? for cable sheaths has been agreed upon by Sweden. The U.S.S.R. figure is much larger; this is a current density of 1.5 »A/ cm? for armoured cables and 7.5 A/cm? for steel pipe lines. Jt is clear that the acceptable figure, in any particular case, largely settles the distances from cables and equipment at which the electrodes are to be placed to minimise excessive corrosion. In the case of submarine cable transmission with sea-return, there will be little effect on land-based equipment because of the current being largely confined to the sea. f There is, however, the question of submerged communication and power cables. The Gotland scheme calculations showed that with a load current of 200 amp and a cable diameter will of 7 cm, the minimum distance between either electrode and the cable should be 9 km in . order to reduce the leakage current in the cable sheath to 0.1 4.A/cm?; a distance of 10 km | was accordingly provided.915) For similar reasons electrodes should be situated away { from dockyards and harbours. ions In the case of overhead line transmission there must also be adequate distances between rf: the electrodes and the line in the case’of earth return. The return current in this case will { tend. to enter through the metal of poles and towers and flow through the earth wire, if ion provided, to leave at other towers where corrosion will result. This is not as likely a source ge of damage as the case of cable sheaths which are particularly susceptible and should be insulated. Whe In the case of the double-conductor line with earthed mid-point, the temporary use of uyit earth return allows a much higher figure of leakage current density to be used. The figure iB. of 0.1 A/cm? can, in this case, be safely increased to 2 or 3 pA/cm? and more if a limit be fe. set to the maximum number of days per annum on which the earth will carry current. This by: type of specification will greatly reduce the distances required between electrodes and : converter stations. Corrosion of cable sheaths, pipe-lines and some other equipment can be avoided by the provision of “cathodic protection” arrangements.410 41) A counter-directional current is impressed, from an external d.c. source, on the metal surfaces subject to corrosion, thereby preventing any outflow of leakage current from the metal. The usual type of arrangement is to place anodes opposite and along the structure to be protected, the structure itself behaving as the cathode. 9.6 Effect on the horizontal component of the earth’s magnetic field (74) 9.6.1 Magnetic field change and compass error 3 The possible changes in compass readings, produced by d.c. flowing in cable(s) on the et sea-bed is of some importance to shipping, particularly in restricted waters with dense traffic such as the Straits of Dover. The direction of magnetising force, due to direct current in the cable, is at right-angles to the cable and the resulting magnetic field would be substantially cancelled if the return current was through an adjacent cable. This is not the case when the sea is used as a conductor since the return current is then distributed. Figure 9.5(a) shows the position of a cable at a depth S, carrying a current J,. At any point, the magnetising force is perpendicular to the cable length and to the line joining that point to the cable centre, as shown by such vectors as H,, H,, H, and H, at points A, B, C, and D, respectively. The component of force causing magnetic compass variation at any each Cornea Ca oc “ a — seats ata aE dima a A DE ease - Figure 9.5 Deflection error in the compass ne edle of the ships due to current in the d.c. submarine cable Lia a COMPASS ERROR 14] point is parallel to the surface and is clearly a maximum at point A where it has a value H,. At point C, a distance x from A, the horizontal component H,=H,(S]S,).cos 0 S Ss eT TS S2 S24x2 1 "TFC ee : See eel el | (9.1) 500.S © [1+(x/S)?] where /, is in amps and x and S are in metres. Thus the greater the depth of the cable, the less the field strength; the field strength also decreases roughly as the inverse square of the surface distance x from the cable. This is naturally true for points on both sides of the cable, and if the current through the cable is reversed then so also will the direction of the magnetic field due to it. The maximum deflection of the compass needle from the N-S direction will occur when the magnetic force due to the cable is in the E-W direction, i.e., the cable is in the N-S direc- tion. This is shown in Figure 9.5 (b) and the maximum deviation Am is given by: ERAN OS ante I fo bd alolle e's dlele ed sialule bet-alalek dh clatele't Uuelp blah a dele p lf elelh bea oll (9.2) where H, is the horizontal component of the earth’s maguetic field.* If the direction of the cable makes an angle A with the N-S direction, the component of its field in the E-W direction is given by: H=H,,. cos A and the deviation A is given by: tanA=(H,/H,) . cos X Substituting in this expression from equation 35, and expressing H, in terms of the cable current and the distance of the cable below the surface: I, 1 500. SH L+AG/S) ccc cece ccc ee ence cence (9.3) Jn the Gotland scheme the cable lies about 11° from the E-} direction and, according to the more recently. published information,@1 the maximum deflectional error is only likely to be 0.3° with 200 amp d.c._ In the English Channel, however, the cable carrying 800 amp (assuming earth return, although this will not be the operating condition since two cables are proposed) from Dungeness to Boulogne will lie at about 35° to the N-E direction, and hence the resulting compass deviation will be of more significance if emergency sea-return operation is to be considered: if the cable is assumed to lie at a depth of 25 fathoms (46 m), the angular deviation of a ship's compass directly over the cable would be 7°, approximately. It should be noted that this deviation diminishes rapidly with distance on the surface from points directly above the cable and, in this case, would have been reduced to 1° 50’ at a =H,. tan A = * In the English Channel He is approximately 0.19 gauss bola Gur a 142 USE OF EARTH AND/OR SEA RETURN point 100 m away. Clearly the only deviation of significance to the course of a ship would occur if the ship was proceeding continually within a very narrow zone along the cable route, 9.6.2 Methods of reducing the compass error One method is to pass the return current through the sheath or an adjacent cable. To use the sheath for such a purpose, it must be insulated from the sea for a voltage at least equal to that caused by fault current. If the insulation is not provided, current will leave the sheath at one end and flow in sea paths to return to the sheath at the other end; further- more the sheath will be heavily corroded at the positive end. The sheath, and hence the system, can be earthed at one point only, and the cable would have to be specially designed to allow for the losses in the sheath. . Ifa second cable is used, then a double-conductor, mid-point earthed system is desirable. The main objection to the use of a second cable is the practical one of raising the correct cable in the event of fault if the two are laid side by side; 500 metres is regarded as adequate clearance for successful grappling. If the two cables are placed at a distance d metres from each other, then the maximum deflection An Will be at a point directly above one of the cables and will be given by: tan A = cosA } 1 l ™ “500.5. H," [ Hal = [EA] “300.7, °S TH? Fee eee eee eee eee eee ee and the percentage cancellation of magnetic field at a point directly above one of the cables will be at: , 1 Ne=s +(d/S2 Thus for the same depth of 46 m and a distance between the cables of 500 m, the per- centage cancellation will be approximately 0.8 per cent and is negligible. A ship crossing both cables will experience the full angular deviation in one direction and then the same angular deviation in the other sense as it crosses the second cable. Clearly the best solution, for a double-conductor system with mid-point earthed, is to provide a twin-core cable, or two cables laid close to cach other. In the case of a twin-core cable, no use of temporary sea return can be made in the event of a fault on one core: in the case of two cables, sea return can be used excepting when the faulty cable is being subjected to grappling or other lifting operations. In the Channel project, it has been decided to lay two single-core cables simul- taneously with a very close spacing, approximately 10 feet. The laying of cables so close to each other is a difficult operation but successful experiments have recently been carried out in the vicinity of the proposed power link. - 100 percent 2.0.0... cece cece cece ececeeeeee (9.5) i A NE Stead ee rele ee aha aa a ce ep ore rt te [ { vould { CHAPTER 10 Harmonics ome leave References: (12) (28) (32) (36) (72) (75) (82) to (109) inclusive, (115) (118) to (124) inclusive, (170) Upeg- 7 i 10.1 Harmonics on the a.c. side of a converter gue The connection of a rectifier or an invertor to an a.c. system is entirely different from that of any other type of load. The mercury arc converter is a switching apparatus, the current waveshape on the a.c. side being discontinuous but, due to the large inductance on the d.c. side, having a flat top in the conducting regions. The analysis of this periodic, non- sinusoidal wave of current is a convenient and simple starting point for the somewhat complex question of harmonics. — juate 2 6 eo a 10.1.1 Current harmonics The problem is the basic one of dividing the periodic, non-sinusoidal current wave into a fundamental sinusoidal component upon which is superimposed harmonics of various orders. Complexity enters the problem on account of the different possible transformer connections, and the degree of assumption which is permitted on account of the presence of circuit elements and the control system. Ss = 10.1.1.1 CURRENT HARMONICS NEGLECTING COMMUTATING REACTANCE AND ASSUMING PERFECTLY SMOOTHED D.C. (a) Star/star connected transformer Figure 10.1 (a) shows the wave-shape of the a.c. line current for the case of a single bridge converter, assuming unity transformer ratio with star/star connection, zero commutation angle and perfectly smoothed d.c. Analysis of this wave shape is obtained from the Fourier series: Ss per- a i=a,+ 2 [a,.cos not-+b, sin nol] n=] Then taking ¥—X as the zero axis, me _2 1, [sin = in| it an + Jy | sin > tin z and b,=0 The terms in brackets become zero for n=2, 3, 4, 6, 8, 9, 10,... etc. (i.e., for all even terms, the third term, and terms that are a multiple of three), and becomes-|:4/3 for n=1, 5, 7, 11, 13,... ete. The solution is thus: _2V3 1 1 1 is 7, [cos olny . COS Sewt-+=. cos Twf—— .cos Nwf+— .cos wt.. ‘ (10.1) 7 7 ll 13 weakens ae sa ewe les 144 MNARMONICS > x “20-0 4msen aT 3 3 3 5 +17 | | Timi nina: i 3T3 T3134 Ly i a Ca A ' ly 6 phase — : ‘ T\| a7 ' 1 # 6)/6 ; paris 3 / 38 ; CE LEO [als 25 @y | | Le 6 ph ' Phase J ay hyfod3 9: ; Poh Lip EEF ©! 933 Po | Tae | | | /2 phase . ' heme : ! bert art pease I Ty, i Z IZ -Il, @ ny a | Val] 03 d @) x Figure 10.1 (a) () () (d) Alternating current wave shape of a bridge converter; commutation reactance neglected The harmonics are thus all of the order 6k-+1, where k is any integer. The r.m.s. value of ariy term of the series is given by: Tono= [GED te V6 hie lhe (10.2) 2 V2 nt Putting n=1, to obtain the tm.s. value of the fundamental: Weert T= O.7B Tyee, (10.3) - i compared to the total r.m.s. current of I= ;- M0816 My ee, (3.2) pecan —_ me CURRENT HARMONICS ON THE A.C. SIDE 145 where J,,) denotes the r.m.s. value of a harmonic of order n, and Jy) denotes the r.m.s. value of the fundamental; Tay. and Joyo denote the p articular values of the fundamental and harmonic of order n, Tespectively, at y=0, From equations (10.2) and (10.3), it may be seen that the r.m.s. value of the n‘# I Cy aii harmonic (L) Delta/star connected transformer If a delta/star (4/3:1 winding ratio) transformer Counection is adopted, the second voltages are displaced by 30° with res; pect to the corresponding voltages in nection. The line current will now be as shown in Figure 10.1 (0). same axis X—X, the Fourier analysis is now: 2/3 1 1 1 i=—.L, [eos ett s . COS sar—; = COs Tat—> +cos Ilwt+t ary star/star con- With reference to the 13 Equation (10.5) is similar to equation (10.1) except that I 19th, 29th, etc., orders are of different sign. EM I3wt+.. | harmonics of the 5th, 7th, 17th, (c) Two bridge units together When two bridge units of equal capacity are working together in ei the one with a delta/star transformer of ratio 2. +/3:1 and the other former of ratio 2:1, then it is clear that the resultant instantaneous c average sum of equations (10.1) and (10.5). ther series or parallel, With a star/star trans- urrent is given by the ie, j= 1 7 Ty [cos wa + COS Hors + COS 13wt— 1 ~3° Cos 23wt+ .. il Equation (10.6) Tepresents the output of a 12-phase converter and h shown in Figure 10.1 (c). Harmonics of the orders Sth, 7th, 17th, 19th ..., which do not enter the system, will circulate between the two bridges; some fraction of these harmonics may, however, enter the system if there is unbalance bety veen the two bridges. In general the order of the current ha as the wave shape tmonics introduced into the system by a p-phase converter will be MERDET ooo cc eeceesee sc. (10.7) where n is the order of the harmonic and k is an integer. P= 6, for 1 bridge p=12, for 2 bridges 30° displaced from each other P=24, for 4 bridges 15° displaced from each other, etc. 10.1.1.2) current HARMONICS, TAKING COMMU ASSUMING SMOOTH D.C. The calculations in 10.1.1.1 took no account of commut TATION REACTANCE INTO ACCOUNT BUT ation angle y; if this is considered, reine — an _ a ft nn et ni SS teed wD Sit Shwe PAE RR aac ln aL A aL Ul 146 HARMONICS i oy oC+ 4 2m Figure 10.2 (a) (6) Current and voltage wave shape ofa bridge rectifier; commutation reactance taken into account the current wave shape will be as shown in Figure 10.2 (b), and may be divided into three parts as in Section 3.9, but taking x—x as the zero axis. ; COS a—cos wf Le, L=[y. Iie ) [cos a—cos (a+y)]’ where a<wt<(a-+-y). . 1 2a 3 3 i,=I,, where (in<or<(F te jh ty ROS A=COS (wt 2H] ee INI ae tL lala —————,, wh sta wt<{~+a vd" “Teos a—cos (a+y)] 3 ) 3 x Fourier analysis of a symmetrical full wave may be simplified by analysing a half-wave and then multiplying cach of the factors a, and b, by 2, and taking n=5, 7,9, 11... ete.: this procedure is complicated. Calverley"70 has obtained a practical solution using a digital computer, by plotting the curves of the harmonic currents /, ()» 48 percentages of the fundamental Tay, against y for different values of a. A precise value for the fundamental current Z1), and hence the har- monic currents also, can be obtained from a procedure given later in this section [equations 10.13 (a) and (4)]; as an approximation, however, Za) may be taken to be equal to Ia, (equation 10.3). Figures 10.3 to 10.10, inclusive, illustr rate the curves of harmonic currents and correspond to the Sth, 7th, 11th, 13th, 17th, 19th, 23rd and 25th harmonic, respectively; other curves for { CURRENT HARMONICS ON THE A.C. SIDE 147 20 oo Ss Ss u DH Vaasa ~ “WN Ca o % of fundamentdl Ly S S oS eae 4 WD iree Ak | + ate 0 /0 20 30 40 Angle of overlap ¥ (degrees) Figure 10.3 Variation of 5th harmonic current in relation to angle of delay and overlap = z harmonics up to and including the 61st are extant.7 _]t may be verified from these figures that the effect on the current harmonics of changes in the angles a and y, are as follows: (i) As y increases, the magnitudes of the hatmonics decrease, but with the higher orders decreasing more rapidly than the lower oncs. (ii) The rate of reduction of harmonics increases as y increases up to a certain limit. (iii) Each.-harmonic decreases to a minimum at an angle y=360°/,, and then rises slightly, = ay 3 thereafter. nk (iv), With a constant angle y, the changes in the various harmonics for different values for of a, js small. a ecressercrncmc <7 eee reer ewe earper ecm en ern pemyerrerpame” - t a i tt Pena na hah eM a RRT Ns Sa oT att Tel aa Call th 6 coined 148 HARMONICS NS S ——}. ~ 4y) RG =7. | | B /9| , // \ | i i 9 NN 8 7 6 5 0 &=30 az 20° ND a=/0 [7 a8 2 % of fundamental =O — 10 20 «30S Angle of overlap W( degrees) Figure 10.4 Variation of 7th harmonic current in relation to angle of delay and overlap (v) For constant current, when the angle a is increased, angle y is reduced and harmonics tend to increase and approach the highest values at y=0. In no case, however, do the harmonics exceed the values given by = Tayo v6 Jono retest eeeeeeees (10.4) Any harmonic which is present with any particul magnitude as when it is present in a six-ph changed, e.g., the 17th harmonic, given by the converter has six or twelve phases. ar number of phases has the same ase converter, if all other conditions are un- Figure 10.7, is of the same magnitude whether eet ata en anager CURRENT HARMONICS ON THE A.C. SIDE 149 S A QA~ © © AN ly as @ % of fundamental Ly) QS ~ NW Q | /O 20 30 40 Angle of overlap ¥ (degrees) Figure 10.5 Variation of 11th harmonic current in relation to angle of delay and overlap The fundamental component of the current wave-shape, taking y into account, can be found by equating powers on the d.c. and a.c. sides; from Chapter 3, the d.c. power 4 Py= Vy. Ty iE _3V2 E [cos a+cos (a+y)]. I, de eis saa Active a.c. power, P,=1/3.E. Taya on the assumption that the a.c. voltage is purely sinusoidal and Tayg is the active (in-phase) component of the fundamental current. Neglecting losses and equating, P,=P, 6 cos a+cos (a+ whence, ian I. (a+) 7 — | aa 150 HARMONICS oF fundamental Ly ) % oP PY S$ Dan waos ds ads a ene Re Angle of overlap 7 (degrees) Figure 10.6 Variation of 13th harmonic current in relation to angle of delay and overlap cos a+cos (a+y) M9» ——— 2 Pe ornesl eeasrent erect eee LULU (10.9) where Kayg= aes (ay) The reactive power on the a.c. side, which is introduced b angle, may be found from ai y phase shift and commutation n equation of the form: tte cece ceneseesesevace (10.10) where i is made up of three ° parts and is given by equation (10.8), and e, is the quadrature component of the correspon: ding instantaneous Voltage. As may be seen from the voltage Serer UTE ome ee 7 f [ : CURRENT HARMONICS ON THE A.C. SIDE 151 S D + J u _ A Q Figure 10.7 Variation of 17th harmonic current in relation to angle of delay 4,080 % of fundemental ly) [ and overlap : 2 / 3 L 0} } 1. O /O 270 30 40 E Angle of overlap U (degrees) 0.9) / 3 curve of Figure 10.2 (a), the instantaneous voltage between phases R and Y, taking Y—X as the reference axis, is given by: e=/2.E.cos (wt—n/6) “.ej=V2.E. sin (wt—z/6) By substitution, the solution of equation (10.10) is:as follows: P 34/2 [?xtsin 2a—sin 2(a+ | =— © ET | "J... 1 : M4 4[cos a—cos (a+ y)] re 7 Also P,= 3. E. hy, where /1), is-the r.m.s. reactive component of the fundamental current. bs cn Wb ae. i a mM AS is Mas Seiler he unease 152 HARMONICS Figure 10.8 Variation of 19th harmonic current in relation to angle of delay and overlap 4g 0s 0 % of fundamental! 4) Angle of overlap V (degrees) V6.1, [te 2a—sin 2(a+ | 4[cos a—cos (a+y)] ; Seda cage ee fos a slelelaFfols ostele tofcieetala petard Sea ee (10.12) It follows that the value of the r.m.s. fundamental current for any angle a and y will be given by: . Iay,= (Dr 7 Zay=V(Payat Pay) Substituting from equations (10.9) and (10.12): 7 Tay= Tayo» V(K aya t Ke ayy) eee eee [10.13 (a)] V([cos 2a—cos 2(a-+y)]}?+[2y+sin 2a—sin 2(a+y)]2)...... [10.13 (5)] ie, Iay=Lay - Le, Iay=Layo 4. [cos a—cos (a+y)] a f t ‘ { fe me tt ennai ae St ta celal thai i nenlai acaini ain i la lebila sein ses nl, wae CURRENT HARMONICS ON THE A.C. SIDE 153 NH iP uw NS ~~ Figure 10.9 Variation of 23rd harmonic current in relation to angle of delay and overlap Ts asa % of fundamento/ ly ) S OT /0 20 30 40 Angle of overlap T (degrees) Calverley has plotted the factors Ka), [equation (10.9)] and Kg, [equation (10.12)] against a for different values of y and using these curves, shown in Figure 10.11, the r.m.s. value of the fundamental can readily be found from equation [10.13 (a)]. These curves also give the active and reactive power for any angles a and y:‘32) Pi=V3.E. ayo» Kaya (10.14) P,=V3.E. Ia. Kay (10.15) The r.m.s. value of Ja), for different values of a can be obtained from the components shown in Figure 10.11. Jt can then be seen that the value for Tq) at a=0, decreases as y increases. Also, for all practical values of y, Tay decreases as y increases independently of a; the decrease is approximately proportional to y, but a few calculations using equation (10.13) and Figure 10.11 show that the change is less than 1 per cent within a practical range of operation. Thus it can_be safely assumed that the factor V/(Kay®,+Ky*,) [in equation 10.43 (a@)] is equal to 1. The implication is therefore that if K(1, (the active rT pees eee a ee rene ER RE EE Re A PC TE wo fat a pam, a ene ead ites ae ta RE Ne i tie ce halal 154 HARMONICS SPM | ir 45 1=25 8 [ “S40 > g 35 8 8 3.0 SS 254+- SS | Figure 10.10 Variation of 2.0. _I\ 2sth Jarmonic current in lation to a dela x | Sag mt er o 75 8 +0 20° LOS SN \ G=/0? @&=20? oO} 230° O /O 20 3O 40 Commutation angle T power output) is known, then Kqay, (the reactive power requirement) may readily be cal- culated from: Kay, ~V (1 —Kay?.) Since Vs=V,. Sarees (Fy) I tedel.. (3.11) and from equation 10.9: Vi=V, . Kayg the curves for Kay, in Figure 10.11 will also give the d.c. output voltage. 10.1.1.3 Power factor Equating a.c. and d.c. power under ideal conditions: VM=V3-E. Tayg and when a and y are taken into account: cos a+cos (a+y) Vi, . —7 v3 -E.Iyy.cos $ . : moterarinegy \ t Bike ~~ vad eas = CURRENT HARMONICS ON TIE A.C. SIDE 155 If Ja, is assumed to be equal to Jay, then comparing the above two equations: cos a-+cos (a+), 2 as was taken to be the case in Chapter 3 (equation 3.12). A more accurate value of the power factor angle ¢ may be found from: cos ¢= P, Tey, Equation 10.9 2y+sin 2a—sin 2(a+-y) COS 2a—cOs app) Tec reteteeeees eee eeees (10.16) Curves (iii) of Figure 10.11 give values of ¢ against a for different values of y. Alter- natively the power factor can be obtained directly from — aa V(Payat Pay) , ie. tan d= (10.17) bird ° 9 © 8 oo e Power factor angle ? (degrees) g ° & = 8 ° ay Kaya andi Kye nn > Ss sss & 8 8 0-6 08 b 10 pL Angle of delay ox (degrees) ‘Figure 10.11 Variation of power component, reactive component and phase angle of fundamental compo- nent of secondary current, with angle of delay and overlap [K(1)2 on the diagram should read K(1)r] 156 HARMONICS All the curves of Figure 10.11 have been extended into the region of invertor operation by taking values of a up to 180°, where a=7—Bf. For purposes of practical estimation, a good rule for determining 4, which may be checked from Figure 10.11, is: ¢xa+2y/3, 0<a<30° graty/2, 30°<a<90° In rectifier literature and practice, when the power factor term, cos ¢, takes fundamental frequency relationships only into account, it is known as the displacement factor A. Thus, for practical purposes A=cos ¢. A further and more accurate determination of the power factor can be obtained by aking the current harmonies {,) and the magnetising current of the transformer J,, into account; this is called the “total power factor”. I Thus, total power factors ——— "Ma 10.18 Hl Vlayrat Cay tn)? + 2 Toy?) ; ) The ratio (total power factor)/A is frequently called the ‘‘distortion factor”. 10.1.1.4 CURRENT HARMONICS NEGLECTING COMMUTATING REACTANCE AND WITH ZERO INDUCTANCE In the derivations above, zhe d.c. was assumed to be perfectly smooth (infinite inductance), but if it is assumed that the inductance is zero the wave shape for zero delay and zero commutation angle will be as shown in Figure 10.1 (d). The Fourier analysis of such a wave shape will be :(118) 2/3 i=2¥3 - [,{cos wt—0.226 cos Swt4-0.113 cos Twt—0.091 cos 11wt+-0.065 cos 13wt— 7 7 —0.057 cos 17wt+0.045 cos 19wt— ...J oe. eee ee cee (10.19) By comparing equations (10.19) and (10.1), it may be seen that Jy), is the same, Is ot I [72]: 100 per cent is increased from 20 per cent to 22.6 per cent [ ma 100 per Qjot * (Qo. cent is approximately the same, whilst all other harmonics are reduced to some extent. In practice, however, a substaztial amount of smoothing is present and the departures from the values of harmonics given by Figures 10.3 to 10.10, inclusive, are very small. Calcula- tions taking into account specific values of d.c. smoothing inductance are very complex and are beyond the scope of this book. 10.1.2 Current and voltage harmonics in the a.c. system The current harmonics discussed above enter the a.c. system and are distributed through- out it in accordance with the network impedances. Consider a simple case of a converter of rating P connected to a purely inductive system, having a short-circuit capacity P, at a voltage E. The converter can be considered as a current generator which feeds a constant x harmonic current of J) into the system reactance X). Thus the percentage voltage harmonic produced at the connection of the system and the converter wiil be: Tuy + OV y= Xe 1 100% Jee 1 tealsledcteues de llbe (10.20) - Be ee Ut [ [ sration ar tcietlinpengnepsenipasenenyt ; . sole : : . . : - de a nae aera encanta ececeene cairo CURRENT AND VOLTAGE HARMONICS IN THE A.C. SYSTEM 157 The system reactance at fundamental frequency is Xyay=E?/P, or in terms of the percentage voltage drop %Vay=[PIP.]. 100% 2. ee cece cece cece eens (10.21) Clearly, the x harmonic reactance, Xu Will be u times Yq), and since the n harmonic current is 1/n of the fundamental current, assuming perfectly smooth d.c. and neglecting y; thus from equations (10.20) and (10.21): : I 1 Tay. X, Wy= aX Be 100-= OO, 100 ° P 0, UV ay=B WOOK eee eeeee cece eee ee eee eee (10.22) s Thus, at the junction of the a.c. system and the converter, the percentage voltage drop of any harmonic order is equal to the fundamental frequency voltage drop. Jn practice, due to the overlap angle y, the actual value of the n harmonic current will be less than Jain, especially in the case of the higher ones; there will be corresponding reductions in the values of Vi,). Equation 10.22, although a simplification, indicates that in order to keep the voltage harmonics below a given limit, a limit must be set to the invertor rating in relation to the short-circuit capacity of the system. It follows that since h.v.d.c. converters will generally be of large power rating, some measures will have to be taken to reduce the current har- “monics entering the system (Section 10.1.4). In the discussion above, the capacitances of the lines, cables, power factor correcting capacitors, etc., have been neglected but in practice have great influence on the harmonics since, with increasing frequency, the capacitive reactance becomes predominant. The general result is that some harmonics, at whose frequencies resonance occurs, will be greatly amplified whereas others may be attenuated or suppressed. Consider an extension of the simple example above, in which the converter, of rating P, has a power factor correcting capacitor of rating P, at a voltage E [Figure 10.12 (a)]. Assuming once again that the system is purely inductive, the n“ harmonic reactance of 2 the system will be Pp and that of the capacitor will be oP" The equivalent circuit is n. s shown in Figure 10.12 (6), in which the converter is considered as a current gencrator feeding the n™ harmonic current into a parallel circuit consisting of the system reactance and the reactance of the capacitor. In such a circuit, resonance can occur at one of the harmonics supplied by the converter. c For res' ee = ance —— =—_ resonan P, nP, ! va pal: Sle het elelh b avila sa sleet dela bd elole ed Hale Seles a ola htm alelh bo (10.23) n2 The consequence of this resonating circuit, which is fed by a constant harmonic current the value of which depends on the converter rating, will be that a very high voltage will be M 9 Seen eter een menace 158 HARMONICS f4 Convertor - i fe pp yi n aie generator Figure 10.12 (a) (6) Simplified circuit for possible parallel resonance between system and power capacitor at harmonic frequencies Also in addition to the harmonic current supplied by the converter, a larger harmonic current, which may be much more than converter current, will circulate between the system and capacitor, heavily loading both of them. Due to the high harmonic voltage, the wave shape at the converter terminals will be greatly distorted, and thereby affect its performance. Furthermore, under these circumstances the calculations of the converter characteristics, which are based upon an undistorted alternating voltage at the converter terminals, will no longer hold good, & Equation 10.23 gives a rough estimate of the relative static capacitor rating, at the funda- f mental frequency, at which resonance can occur. * for n=5 P.=0.04P, [ ; for n=7 P.=0.02P,, ete. ! This is a simple example of resonance occurri : converter. Harmonics may well excite other Pp E The possible combinations are so many, and dependent on loading conditions, that it is extremely difficult to generalise; some examples and comments may, however, be of value. i Induction motor load together with power factor correcting capacitors form a parallel [: [ set up across the system supplied by the converter. ng in the immediate neighbourhood of the arts of the connected system to resonance. resonant circuit; when the load is disconnected, the capacitors form a series resonant circuit with the transformer inductance. : aera DISTURBANCES IN NEIGHBOURING COMMUNICATION SYSTEMS 159 A long unloaded transmission line connected through a transformer is another likely series resonant circuit. Such serics resonant circuits shunt large harmonic currents from the rest of the system, thereby becoming overloaded and producing high voltage drops across their component inductive and capacitive reactances. Harmonics may also cause undesirable additional losses in the damper or main rotor windings of synchronous and induction machines. Heating due to this cause is generally restricted to harmonics up to the 7th order; most a.c. systems have resonant frequencies between the Sth and 11th harmonics, and the effect of system capacitance is thus to shunt those of higher order. Dificulties may also arise because of propagation phenomena. At the higher harmonic frequencies, transmission lines of quite average length become electrically very long; for example, at 25th harmonic frequency (1,250 c/s on a 50 c/s basis), a line 37.2 miles (60 km) long represents a quarter wave-length. Consider a transmission line connecting a converter to a system; the transmission line may be a quarter-wavelength or odd multiples of a quarter-wavelength. Suppose also that the system presents a very low impedance to the harmonic due to, say, a series resonant circuit in the neighbourhood of the connecting point between the system and the line. If the damping of the line is small, it will present a very high impedance to the harmonic current entering it at the converter end and a high harmonic voltage will be set up there, assuming that no capacitors are provided. Also the harmonic current entering the system, and at any current anti-node in the line, may be much more than the converter harmonic current. Again, if the line is a half-wavelength, or even multiples of a half-wavelength, and the system presents either an open-circuit or a high-impedance to the harmonic, then a very large harmonic voltage will appear at both the system and converter ends of the line and any voltage anti-node in the line. Furthermore, the current at the current anti-nodes in the line may be several times the corresponding harmonic current of the converter. If the damping of the line is high (cables tend to have a high damping effect), the harmonic currents may be absorbed by the line and not enter the system at all. The Principle of Superposition can be used for studying harmonics in these lines and { systems, a circuit being established for each harmonic frequency. Each converter is represented by a harmonic current generator in series with its own internal impedance and interconnected with lines, loads, transformers, etc., each represented as an equivalent—z network for that particular harmonic frequency. Remote sections of the network may be Tepresented by lumped parameters. All capacitance will tend to be significant and must be taken into account. Problems of this type are best studied on a model network,“ and it is preferable to represent neighbouring transmission lines as a number of sections in order to study the effect of travelling waves.” { ' i 10.1.3. Disturbances created in neighbouring communication systems Disturbances in communication systems due to harmonics in a power system may be through the medium of either electromagnetic or electric induction. The communication systems may,be classified readily into: ol os Mloae C3 Uren a Bieedd 160 HARMONICS (0) Nature of electric induction - Figure 10.13 (a) (6) Nature of induction in communication conductor from power conductor — { , Seba, nee cain iS btn into i anes Ash tr ecient aaa abs DISTURBANCES IN NEIGHBOURING COMMUNICATION SYSTEMS 161 (i) Telephone circuits. (ii) Telegraph circuits. (iii) Signal circuits. Telegraph and signal circuits usually carry currents whose main components have fre- quencies of Iess than 300 c/s, and hence they tend to be affected only by the fundamental frequency and the lower order harmonics. Telephone circuits, on the other hand, carry currents of frequencies in the audio range lying between 100 c/s and 4 ke/s, although those below, say, 250 c/s are relatively unimportant. Telephone circuits are the most sensitive to outside interference from power lines since their small currents are directly representative of speech and induced harmonics may render it unintelligible. 10.1.3.1 ELECTROMAGNETIC INDUCTION The voltage induced in the communication circuit by the magnetic field of the power circuit depends upon the magnitude of the power circuit current, frequency, length of parallel run of the power and communication circuits, and the mutual inductance. Figure 10.13 (a) shows the nature of the magnetic field induction.23) Z, and Z, are the finite impedances to earth of the communication circuit beyond the region of influence of the magnetic ficld of the powe: conductor. If / is the length of parallel run of the two circuits and Af is the mutual inductance per unit length, then the induced voltage due to the n‘ order harmonic current will be: Eqye amo ML My cece ccc cece eee e ees (10.24) The current circulating through Z, and Z, will be: rE! Nyame wn we eee ev cence eceuneevneees 10.25, Ze (10.25) {feither Z, or Z, is disconnected, the circulating current due to magnetic induction will be zero. The voltage induced in the communication circuit can clearly be represented by a voltage Eu in series with the conductor as shown. 10.1.3.2 ELECTRIC INDUCTION The electric field surrounding a power conductor will induce electric charge in any neighbouring communication circuit; the amount of charge varies directly with the voltage of the power conductor and depends on the capacitances of the communication conductors to the power conductors and to earth. The nature of the electric induction is shown by Figure 10.13 ().422) It can be seen from the diagram that the effect of electric induction may be simulated by means of a generator of voltage Eq, the n™ harmonic voltage of the powcr line, connected to the communication circuit through a capacitor Cy2. Then, if the capacitance to earth of the communication wire is C2 and Z4=Z,=, then the induced voltage will be: (10.26) If Z, or Zz is zero, or if the communication circuit is short-circuited at any point, the Current through the short-circuit will be py =O Caps Egy cecseveseveneneeeseeeses (10.27) | 162 HARMONICS If it is short-circuited at both ends, the current through each short-circuit will be 0.5K). The actual current through finite values of Z, and Z, may readily be calculated from Figure 10.13 (6). Numerous empirical and theoretical formulae are to be found in the literature, for estab- lishing mutual inductances and capacitances, and for calculating induced voltages for different systems and configurations when factors such as screening, earth conductivity, circuit asymmetry, ete., are taken into account,(75) (76) (77) (78) (80) (3D (123) Only the main aspects of this work can be discussed here. E 10.1.3.3 LONGITUDINAL AND TRANSVERSE VOLTAGES The general effect of electric and clectromagnetic induction from a power to a com- munication circuit is the appearance in each wire of the latter of longitudinal voltages of corresponding frequencies. Since the communication circuit wires are generally spaced asymmetrically with respect to any one power conductor, the nearest communication wire will have slightly higher induced longitudinal voltages. It is the difference between the two longitudinal voltages, one in each communication wire, which circulates harmonic current and acts in the case of a telephone circuit to distort specch directly; this difference voltage is known as transverse voltage and is clearly a small fraction of the total induced longitudinal voltage. Similarly, the other power conductors of a polyphase circuit will induce transverse voltage in a telephone circuit substantial can- cellation occurring in the latter if the polyphase power voltages and currents are balanced; under these circumstances induction from power to telephone circuits may be significant only when the spacing between the two is small. With telegraph circuits (which may be single-wire, earth-return arrangements), the disturbance will be through longitudinal voltage circulating current through the earth. At the same time the disturbing effect is offset by the high operating voltages and currents of telegraph circuits. The usual hazard of longitudinal voltages is that they may be suffi- ciently high during power line earth faults to cause shock to personnel or damage to apparatus, and protection is necessary. [ 10.1.3.4 UNBALANCE OF TELEPHONE CIRCUITS In a telephone cable, the two wires forming a pair are twisted together with very small separation between them, so that the induced longitudinal voltages are equal. In overhead telephone line, transverse induction can be reduced by transposition. Even with such uniformly induced longitudinal voltages, transverse current will flow if the two wires do not have the same series impedance, capacitance and leakage to earth. Unbalance to ° earth is the predominant factor and C.C.I.F. Directives) recommend a maximum per- missible unbalance of this nature of 4 per cent for open wire lines and 1 per cent for cable circuits. Unbalance may, of course, be formed by broken insulators, stray earth con- nection, etc. : Mee 10.1.3.5 UNBALANCE OF POWER CIRCUITS : “The only unbalanced harmonics in a three-phase balanced system are those of triple Ren Shy: igee sta b- for fe nea DISTURBANCES IN NEIGHBOURING COMMUNICATION SYSTEMS 163 orders. Clearly the inductive effects of these unbalanced harmonics are much greater than those of balanced ones in view of the substantial cancelling which takes place in the latter cases; the power line conductors, for the case of triple order harmonics, behave as a zero phase sequence circuit with all line conductors in parallel, earth or the neutral conductor forming the return path. Fortunately, converter loads, being balanced, produce harmon- ics of non-triple orders only, disturbances thus being restricted to these orders. It is only in a balanced system that non-triple order harmonics are balanced and self- neutralising. In an unbalanced system, unbalanced non-triple order harmonics will appear even though they are balanced at the source. Such unbalances in the system are mainly caused by inequalities between the capacitances of the conductors to each other and to earth. This type of asymmetry causes unbalanced capacitive currents, the vector sum of which is finite and takes a zero phase sequence path through the earth. Because of this, a relatively large value of zero phase sequence current in the neighbourhood of earth points is to be expected, compared with the values existing at points remote. . Protection is the major factor affecting earthing, but it should be noted that earth points can otherwise be selected in such a way as to decrease zero phase sequence current whenever there are parallel communication circuits; in practice, close co-ordination between the two sets of requirements is necessary. For example, if the a.c. system neutral is in- sulated (or earthed through a high impedance such as the impedance of a Petersen coil to high order harmonics), the zero-sequence voltage between neutral and earth adjusts itself so that the resultant capacitive current is zero. Since interference from voltage harmonics (electric induction) is generally smaller than from current harmonics, this type of arrangement will cause less communication system disturbance. This practice of treating a.c. system neutrals is not, however, desirable in e.h.v. systems from the point of view of protection. Every effort must be made to reduce the unbalance of the system since the unbalanced zero-sequence harmonics, although smaller in magnitude than the balanced ones, have a greater disturbing effect for the reasons given above. Generally, for a known configura- tion, the degree of unbalance can be calculated and hence the unbalanced harmonics produced by a known amount of balanced harmonics can also be calculated. Transposition of power lines will help to decrease the induction from balanced harmonics, since the communication conductors will then be equally exposed to the harmonic currents in all the three phases; substantial reduction will only be accomplished if the transposition is done frequently. In case of unbalanced harmonics, the transposition of power lines will not decrease the induced voltage directly but will indirectly in the sense that it will decrease the total amount of unbalanced harmonics. This will be so beeause at every transposition point the un- balanced current is phase-shifted by 120° and the total neutral current is greatly reduced; although now small unbalanced currents will be circulating between the transposed scctions. Against this, the transposition of communication lines will reduce the induced voltages due to both balaticed and unbalanced harmonics in the power lines. This is by reducing the transverse voltage by bringing about equal exposure and better balance of the induced longitudinal voltage in the two conductors. oe oe or i HARMONICS > communication cach supporting pole. Trouble may also arise due t ductors. Because of this, cur! by the large balanced harmoni earth and will Produce electror 'y of the earth wire to the agnetically induced in the earth wire This current will circulate through the in the communication lines,79) Power con- rent will be electrom ics of the converter. magnetic disturbance 10.1.3.6 scrEENING EFFECTS Telephone circuits in cables, ing sheath, are completely free from electric induction, if the sheath is grounded at lea st at one point. Electromagnetic induction sets up a longitudinal e.m.f. along the cable sheath which causes a sheath current to flow provided that the she i both ends, and which p conductor within. This Protection increases with increase in frequency; a lead sheath of 200 mm cross-section with no armouring, the percentage is reduced, 79) approximately, as follows: which have a surround induced voltage Frequency, c/s % induced voltage age than unarmoured Ones, » rails, counterpoises, etc.) in ced circuit, have a shielding 'S very close to the conductor ecial shield, including earth- wires, cannot thus be justified in general. In cases where interference is largely caused by unb: an earth wire may reduce the interference up to 50 per cent, depending upon the relative resistances of the earth wire and the earth (determining the proportion of current returning through the earth wire), circuit configurations, distance, ete, alanced harmonics, ication circuit, Telephone thus the Tesponses of both ; they are commonly taken to varies with frequency, 800 c/s is t and the voltage and current (both inducing and induced) of lated to 800 c/s by means of a “weighting factor”, » “@ Corresponding to various harmonic frequen The total sum of translated harmonic Voltages is ¢: > any other frequency is trans- Figure 10.14 shows the weighting cies, 28) alled “equivalent disturbing voltage”, — leon ing — eet atc es he LR a Sale ald ee SS meet eerie ei DISTURBANCES IN NEIGHBOURING COMMUNICATION SYSTEMS 165 and this, when expressed as a percentage of the power frequency line voltage, is known as the “telephone harmonic factor” and is given by 100 . /2(A gy? . Vey?) % aie eee pee eee Equation 10.28 applies also to current by substituting Ty for Vi and J for V. The ratio of magnitude of the speech signal to that of the induced noise voltage (signal/ noise ratio) is the criterion of the disturbance. For simplicity, the C.C.I.F. Directives) have suggested a total weighted noise level, i.e., the r.m.s. total of all induced voltages weighted to 800 c/s, of 2 mV for cables and 5 mV overhead lines. Another limit suggested by C.C.I.F. is that the percentage value of V(g00), as calculated from equation 10.28, should not exceed 2 per cent at any point on the power system, and the suggestion has even been made, in the light of further investigations, that this figure (10.28) 3 Ne) (r) cf. 800 An) 7d Arn teogg > 99990N 8 6 + 4 a t Ou Hl 13 1719 9325 29 3/ Order of harmonic Figure 10.14 Weighting factor i) CHE (rr * 166 HARMONICS BYR () \ METHODS OF REDUCING HARMONICS 167 should be reduced to I percent. In this case, however, the weighting factor is to be taken as Aww -f 8 instead of Aq; the general term for the percentage Vgco) in this case is the “telephone interference factor’’. Quite clearly in most cases the h.v.d.c. plant will exceed the size indicated by this limit set by interference and methods must, therefore, be adopted to limit the harmonics entering the system. On the other hand, it should be stressed that the application of the limits set up above depend on the circumstances; they may be unnecessarily severe if there are no telephone lines in the vicinity, or may not be adequate if they are present. Extensive experiments in the Soviet Union have shown!) that for cases of separations of 30 m for 35 kV and 110 kV transmission lines, 50 m for 220 kV lines, and 60 m for 400 kV lines, electric disturbances are negligible and only electromagnetic disturbances need to be taken into account. It was also clear that for large separations between the power and communication lines (more than 200m), the only interference in the telephone circuit was due to zero sequence current. 10.1.4 Methods of reducing harmonics on the a.c. side 10.1.4.1 INCREASE IN NUMBER OF PHASES It is clear that the harmonics entering the system can be greatly reduced by increasing the number of phases at which the converter operates. In an h.y.d.c. project at least two bridges can be expected and they can easily be operated together as a twelve-phase con- verter, by arranging the converter banks with 30° phase shift between them (Chapter 2). But to increase the phase number to more than twelve, it will be necessary to provide phase shifting transformers. To obtain 24-phase operation from four bridges, it is necessary to phase-shift the transformers by 15° from each other. One way of arranging this is shown in Figure 10.15 (a). The secondaries of two transformers, the one with delta primary and the other with a star primary, are given a clockwise (positive) phase-shift of 7.5° and the secondaries of the other two the one with delta primary and the other with a star primary, are given an anti- clockwise (negative) phase-shift of 7.5°. Thus, if the transformers are numbered 1, 2, 3 and 4 as shown, the phase voltages of the secondaries will be as shown by Figure 10.15 (8). One of the phase-shifted windings is shown in Figure 10.13 (c), and from its geometry it can be calculated that, for a phase-shift of 7.5° E,1=0.9162E,; E,o=0.1506E, Ys Eyy-bE=10G68E, «2. csececccesscesscevees (10.29) Thus an extra winding, of 6.68 per cent, will be needed on each secondary. The phase-shifting windings can alternatively be provided in the neutral, or on the primary side. All such arrangements are costly, however. A scheme of 24-phase operation could also be obtained by giving 15° phase-shift to two transformers instead of 7.5° to each of four transformers. If normal transformers are already in existence, a separate phase-shifting auto-transformer is necessary; one type is shown in Figure 10.15 (d). The voltages to the converter transformer can be shifted, — a cr om Ositive or negative direction by con- » B,C’ to the transformer, or Vice ‘ hases could be obtained from eight bridges by Providing 7.5° phase- ages of the adjacent transformers. Even though the phase number was doubled from, Say, twelve to 24, Operation the System may still not be ready t © accept the harmonic Operation for half the rating. ~ The considerable extra e: against that involved in oth related also to the ex In the U.S.A,, the i ial rectifier instal] this type of trouble arises; j increased to 108; Versa, (124) Opcration with 48 pi shift between the volt under emergency S of twelve-phase *pense involved in increa ter methods of reducin increase the Phase number b duce harmonics, gett acted sitet (10.23) A possible, but not desirable, way of exercising control to determine t he rating of the capacitor bank in accord capacity P, so that parallel resonance occurs at lowest converter harmonic; thus for twelve-ph for 4th harmonic frequency, Over parallel resonance is ance with the system short-circuit a frequency much lower than that of the ‘4S¢ Operation, resonance could be arranged clearly and by taking P,as the maxi for any higher harmonic. however, since the minimu System short-circuit level rating so determined will be too large. In any event, because of the stray system capacitances and the Possibilit effects, the system inductive reactance at the n'* harmonic fre be 2X qy and will Probably be very low for some harmonics, Series reactors may be connected in the lines on the system sid will serve to preve i A further Possible method of preventing reso factor correctin 8 capacitors is the Provision of Feactance in series with each ¢. that series resonance is obtai i and not by the converter rati Y of propagation quency cannot be assumed to Furthermore, low-inductance ¢ of the capacitor banks, and Bee ¢ Ss Spot pee 5 e M te a gn @ METHODS OF REDUCING HARMONICS 169 dic. Qc. side side system ea ‘ | = 6 Phase converler Cs Ls C7 (7 Cu Li 5. 7th Nth’ (2) luned shunts for harmonics on Q.c. Side Line Rectifier equipment (6) Low pass filter as provided at Vastervik in Gotland scheme Figure 10.16 (a) (6) Filter circuits for harmonics on a.c. side for six-phase operation, etc. With any higher order harmonic, resonance with the system will be avoided in addition to providing a low impedance path for lower order harmonics. High order harmonics will enter the system, but their magnitude will be small. For six-phase operation, as an alternative, it may be preferable to tune the series circuit to the 6th harmonic in order to shunt the 5th and 7th harmonics effectively, and to tune to the 12th harmonic likewise in a twelve-phase system. If it is desirable to eliminate harmonics of still higher orders from the system, then additional tuned circuit filters must be provided, as shown in Figure 10.16 (a). It must be Stressed that a tuned circuit of this type for a high-order harmonic cannot be provided unless tuned circuits for the lower-orders are also available. Otherwise, there is a likelihood ERRATUM Page 169. Figure 10.16 (b) The three filter connections should each be made to a separate Phase, and not all to one phase, as shown. 170 HARMONICS of a higher order tuned circuit(s) resonating with the system reactance at a lower harmonic frequency. This possibility should not arise and economy introduced if tuned circuits for 6th, 12th, 18th,... harmonic frequencies are provided to shunt the Sth, 7th, 11th, 13th, 17th, 19th, ... harmonics. If Xcqy and X; 1 are the fundamental frequency reactances of the capacitor and inductor, respectively, then for resonance at " harmonic: : 1 Xca ob= S—0F Xyay= at se blletiles blseclee shah (10.30) If £ is the fundamental line voltage, the fundamental voltage across the capacitor, nz Vay=E. MD titties (10.31) the fundamental kVA rating of the capacitor n2 2 wCE2 (+) mma W hla p btold blelolablalald llclaldolela (10.32) the fundamental reactive power supplied by the filter circuit n = WCE —— J occ ccc cece cece 7 w (44) ; (10.33) and the fundamental kVA rating of the reactor Capacitor kVA n = Ck? 2 ee i = w Go (10.34) From equation (10.33), it is clear that by providing a suitable size for the capacitor and a corresponding size for the reactor, the tuned circuit can be designed to supply any amount of reactive power, although the capacitor rating will be somewhat greater than the reactive power which it will supply. E.g., if the value of wCE2 is taken to be 100 per cent, and a=5, then at fundamental frequency the reactive power supplied [from equation (10.33)] : 25 - 100=104.2% (25—1) ihe From equation (10.32), the capacitor rating 25 I =]>-— | . 100=108.59 ls % and from equation (10.34), the reactor rating 108.5 pac =4 39 =35 4.3%. It can be scen that the additional capacitor rating of (108.5—104.2)=4.3 per cent is required for magnetising the reactor. In addition to the fundamental frequency loading, the capacitor and reactor will have to carry the harmonic current for which they are tuned. Capacitors are commonly designed for this application with a voltage capability of 110 per cent of nominal rated voltage and a kVA capability of 135 per cent of nominal rated kVA. This margin permits a capacitor to carry, together with its fundamental current, a Sth harmonic of 150 per cent of fundamental current at 100 per cent rated voltage, or a 5th harmonic of 90 per cent of fundamental current at 110 per cent of rated voltage, and similarly for higher order harmonics,.“20 om Gas ane a ine sa nena ost aint ia Reh VOLTAGE HARMONICS ON THE D.C. SIDE 171 Thus, in the case of the Sth harmonic, if the capacitive power supplied by the filter combination is not less than, say, 10 per cent of the converter rating, the capacitor rating may not be greater than »CE®. For higher order harmonics this minimum limit of reactive power is still less. If more than one tuned circuit is to be provided, each one could be designed to supply some reactive power in the most economical way. If the system is capable of supplying all the required reactive power, then the filters could be designed to carry only their corres- ponding harmonics; the filters would probably then be uneconomic compared with in- creasing the phase number. Another method of avoiding resonance with the system and also shunting the harmonics, would be to provide a low-pass filter designed for a minimum frequency lower than the lowest harmonic. This would absorb a large proportion of all harmonics and could also be designed to provide the required reactive power to the converter. Such an arrangement is shown in Figure 10.16 (6), and has been provided on the mainland station (Véstervik) of the Gotland scheme. It has been designed with a damped resonance circuit for all harmonics above the 4th, and with a predominantly resistive impedance for all higher harmonics. At 50 c/s the attenuation is low, but still sufficient to supply a reactive power of S50 MVAr. It is interesting to know that after providing this filter arrange- ment, it has been found possible to run the mainland station (which consists of two bridges) as an invertor station, without providing compensating reactors (Chapter 3). This is very advantageous and economical.15) 10.1.4.3. FACTORS AFFECTING SYNCHRONOUS MACHINE DESIGN If a synchronous capacitor is provided at the invertor end of the h.v.d.c. system, its structure can be modified to shunt some proportion of the harmonic currents and reduce thereby the proportion of harmonics in the a.c. system. The rotating field produced in the stator by the n harmonic current will induce slip frequency voltages of (n—1) times funda- mental frequency in the rotor; the synchronous machine will thus act as an induction machine to all harmonic frequencies. Advantage may be taken of this by providing low resistance damper (amortisseur) windings on the rotor and building the machine for high speed in order to obtain a low sub-transient reactance.(72) If the synchronous gencrators at the rectifier end of the h.v.d.c. have specifically been built for this purpose, they can be modified in the same manner, the source reactances to the harmonic current will thus be the lowest possible and will serve to prevent harmonic voltage drop with consequent distortion. A combination of synchronous capacitor-static capacitor is to be preferred on technical grounds, however; the synchronous machine can be used for regulation and as an alter- nating voltage source of fixed frequency when the h.yv.d.c. system is running alone, whilst the static capacitors are more effective for shunting harmonics and supplying reactive power, and have very small losses. 10.2 Harmonics on the d.c. side of a converter 10.2.1 Harmonics in the output voltage, before the smoothing reactor On the d.c. side of a converter, the output voltage waveshape is of a form which depends a “a Gaby | 172 HARMONICS upon the number of phases, angle of delay and angle of commutation. The output voltage can be considered as a d.c. voltage on which are superimposed a number of harmonics and can be calculated by Fourier analysis. With zero delay angle and no load, it can be shown that the r.m.s. value of the n" har- monic voltage will be given by, Vino=Vox Gary etic eres (10.35) where n=mp, m being an integer and p the phase number at which the converter is operating [depending upon the number of phase-shifted bridges and method of connection (Section 10.2.2)]. With an angle of delay a and angle of overlap y, the output voltage is as shown in Figure 10.2 and may be divided into three parts. ie., e,= 1/2. E. cos (Z+01) where 0<wt<a e=a[ v2 . Ecos (or-2)4.y2 . Ecos (=) ee (10.36) =V2.E. cos &. cos wt where a<wt<(a+y) and e,=1/2.E.cos («:-2) 7 where (aty)<ot<3 Fourier analysis of equations (10.36) give the cosine and sine terms of the 2 harmonic as: a=ay{[o-0 - cos (n+ 1) (+5) - COS +05] Y Y —(n—1). [cos (n—1) (+5) a D5] Sass (10.37) =v acta Y apy bai {fo 1) . sin (a+1) (+3) = COs (” +05] : y y —(+l). [sn (n—1) (+5) cos (n—I) 31} ee (10.38) The r.m.s. value of the 2 harmonic will be: Vy~ |. ; v[@—pe - cos? (a+ 1] y/2)+(2+1)2 . cos? (f2—1]y/2) N —2(n—1) (n+l). cos ({2+1] y/2) . cos ({n—1] y/2). cos Cnty] beste atestate (10.39) voltage mgnics 5. ( Bs crpsing if on tT 10.36) [ L mae me, ~. - ae. SS ARRRRRE c caae Aliaentln aade ei aa A as i VOLTAGE HARMONICS ON THE D.C. SIDE 173 ASEA of Sweden have provided a set of curves for Vo as a percentage of V, against y for different values of a as shown in Figures 10.17, 10.18 and 10.19, and are for the 6th, 12th and 18th harmonics, respectively. Veyos AS @ percentage of V,, can be found from equation (10.35) and is equal to 4.04 per cent, 0.99 per cent, 0.44 per cent and 0.25 per cent for the 6th, 12th, 18th and 24th harmonics, respectively. 20 SIX ov 7 -— Let ithe 10 gS Sa sl al 5 Meo o+ —— 0° -. fo° 20° 30° 40° Angle of overlap ¥ Figure 10.17 Variation of 6th harmonic voltage in relation to angle of delay and overlap 174 HARMONICS 30 4o° Angle of overlap ¥ Figure 10.18 Variation of 12th harmonic voltage in rel. lation to angle of delay and overlap It may be seen from the curves that for small values of the angle y, the harmonic magni- tudes increase with increase in angle a and that the higher the harmonic, the more rapid is the increase. For a constant angle a, the harmonics decrease (they may increase slightly at first for small angles a) and reach a first minimum at, approximately, y=a/n; for y= a/(n+1) and y=z/(n—1), the harmonics are constant for any angle a; at y2r/n, there is a maximum and at y=3z/n there is a further minimum. ager | | CIRCULATION OF HARMONICS ON THE D.C. SIDE 175 10.2.2 Circulation of harmonics on the d.c. side, with different methods of connecting the bridges The path of the harmonic currents depends upon the method of connecting the bridges (Figure 10.20). The connection of the neutral to earth provides a path for a number of harmonics. When two bridges are connecting in series and displaced by 30° from each other, the harmonics of the orders that are odd multiples of 6 (6th, 18th, 30th... .) will be equal and opposite in phase, whereas harmonics of the orders that are even multiples of 6 (12th, 24th, 36th...) will be in phase. This can be confirmed from the equation: are tan Qaymct Se talelibhasttete bt date t daleleb s (10.41) In where a, and 5, are given by equations 10.37 and 10.38 and 9, is the angle of the n™ har- monic with respect to the zero axis (Figure 10.2). |= 18 Z oO 8 =< \ \\ A NS 90 30 40° Angle of overlap ¥ Figure 10.19 Variation of 18th harmonic voltage in relation to angle of delay and overlap essa rt a ann tt A i aa a NS Na nN as sab asalsi eg 176 HARMONICS Thus, when both bridges are on the same side of earth, cancelled whilst the latter harmonics will sum and circulate rents, the d.c. transmission thereby 10.20 (). However, when the neutral is carthed the former harmonics (odd multiples of the 6th), the two sets of which are equal and opposite in the bridges, become in phase along the earth path. The latter harmonics (even multiples of the 12th), the two sets of which are equal and in phase, will circulate corresponding currents through the two conductors, as shown by Figure 10.20(c). It is clear that the former harmonics will be in phase along the two conductors, whilst the latter harmonics will be in Phasc-opposition along the two conductors. If two bridges are connected in series and are in phase, clearly the system will have a six-phase operation whether the neutral is earthed or not, and all the harmonics will return through the other conductor as shown by Figure 10.20 (d). An arrangement having two bridges in series displaced by 30° and with no earth con- nection, or earth connection at one point only, is obviously the best from the point of view of harmonics, but may be uneconomical. The Channel project will be earthed at one point only: the smoothing reactor will be connected between the two bridges (30° displaced) on each side of neutral, and the mid point of one reactor in one terminal station only will be earthed through an impedance consisting of an earthing reactor and resistor in series. the former harmonics will be corresponding harmonic cur- having a twelve-phase operation as shown in Figure 10.2.3. Harmonics in the d.c. lines) 022) It has been shown in Section 10.2.1 that on the d.c. side, the converter can be considered as a generator of voltage harmonics, which will circulate current harmonies of correspond- ing frequencies in the system; thus the current and voltage harmonics at any point in the d.c. transmission line can be obtained from the equivalent circuit of the system for the n harmonic, shown in Figure 10.21 (a). Here only the n harmonic introduced by the rectifier is considered, but the effect of invertor harmonics can be similarly considered by inserting a voltage generator on the invertor side. Lyg is the total rectifier reactance consisting of d.c. reactor and rectifier internal reactance (resistance can be neglected); similarly, L,, is the corresponding invertor quantity. At harmonic frequencies, a d.c. transmission line must be considered as a line which is electrically long; e.g. for 12th harmonic frequency (on a 50 c/s basis), the line has to have a lengtlr of 77.5 miles (125 km) only to represent a quarter-wavelength. The line can be represented by its equivalent 7 or T network ; referring to Figure 10.21 (6): = ; _., sinh V(Z. ¥) Za=Z,. sinh W=2— Ta cneeoeekees Seeds (10.42) yb Yr_tanh oe Y V(Z. ¥)/2 and 2 sega tanh ZY poaarerec tse (10.43) Z, where 2= fim [Zn [Reson $a ade (10.44) y Y G+jenC 300, 600, 900, /:200-- a w ) 600, 1.200, |800--- 300, 600,900, 1200 -- " + 600 12900 1800 --- ~ Figure 10.20 (a) to (ce) Circulation of harmonics on d.c. side with different methods of connecting the bridges (300, 600, 900, ete., correspond to 6th, 12th, 18th, harmonic respectively) ro : ‘ a amie ok it aint ae a al saisbin a ines iaaitlsin ennai i 178 HARMONICS and is the characteristic impedance of the line. Also, z=n" harmonic line impedance/unit length=Z/I y=n" harmonic line shunt admittance/unit Iength= ¥/1 y=Propagation constant =Ve. VED (Rt font) CtjanO). where a=attenuation constant/unit length B=phase constant/unit length. [Note that the complex number y and the constants a and 8, as used in this Section (10.2.3) only, are the symbols normally used in transmission line theory, and are not related to the same symbols used elsewhere in this book for commutation angle (y), delay angle (a) and advance angle ().] Alternatively, referring to Figure 10.21 (c): Zy_ Be vz. ¥) zn tanh (5 =3° tanh. Teen (10.46) Vz. ¥y2 _Sinh yl | sinh /(Z. ¥) and Y;= Z, =f!Y, Wz. Y) tea sele es salen edsle (10.47) If Z(. is the total impedance of the transmission line and the invertor at the n™ harmonic frequency, viewed from the rectifier end, then the harmonic current Taya through the rectifier reactor, due to the rectifier harmonic voltage Vy, will be: V, _ ) Jor= = Jeong tZgy ees (10.48) Zz, Jonka . cosh WZ, : sinh yl Z,. cosh yl-+-jonl 4, . sinh yl maz, oe hte o tenho ||! || ||| Ll) (10.49) Z.+jon . Ly. tanh yl The voltage harmonic Voyrs immediately after the reactor is given by where Z(,)= Vour=Tayr . Zm : = Vay Taya ion. Lag) ov eee ccccceeees (10.50) Another possible way of calculating Z,, is as follows: If the invertor end is assumed to be purely inductive, it may be represented by an ex- tension of the transmission line (short-circuited at its end) represented by its angle 4 given by jon, - y=are tanh (= 3) tele ed delet deal hat. (10.51) c Since the real term of Z, predominates (equation 10.44) it is the imaginery term of yp which predominates. cual ryaie eS — HARMONICS ON THE D.C. LINE 179 Thus Z,)=Z,. tanh (yl)... 0. eee ee eee eee eee (10.52) =Z,. tanh [al+j(Bl-+yY)] ..........00.00. (10.52) The voltage and current harmonics at any point in the d.c. line may be required. The n‘* harmonic current or voltage at a distance x from the invertor end [Figure 10.21 (a)]. can be found from: Jn, 008 (xt) _cosh [ax +j(Bxt+H] (10.53) Nar cosh (yl-+¥) — cosh [al+j(Bl+y)} i Veye_sinh (yx-+$) _sinh [ax-iy(Px+¥)] Voor sink (FEY) Sin [aA] where Toye Voyr vy and # are given by equations 10.48, 10.50, 10.45 and 10.51, respectively. For simplification, it may be assumed that the impedance of the remote end is infinite, i.c., open-circuit at the remote, or invertor, end. On this assumption, ZoyHZ,.Coth yl vies eeeeveeeeeeeeeeeeees (10.55) Tgyx sinh yx AAee ener (10.54) eo eaelslee esa eee eee mess susie t 10.56 Tqyp Sinh yl ( ) and Vonx _cosh yx seen cence ene eee e eee eeee 10.57. Veyr cosh yl ( ) The value of the damping constant a/ is high for the case of cables and harmonics can be expected to be correspondingly attenuated. Overhead lines, however, have low damping and approximate to lossless lines with high inductive reactance terminations. These high reactance terminations give reflection factors close to unity, and high amplitude standing jonL, : : Jen"ak (approximating to waves of voltage and current can occur. For very high values of . c open-circuit) and if the line length approaches G wavelength) where k is an integer, harmonic current nodes will appear near to both the rectifier and invertor ends. The result of approaching such a condition is serious since the current at the anti-nodes on the d.c. line will be several times the current at the source, i.e., on the rectifier side of the rectifier reactor. The reactor is thus rendered incffectivé, the harmonic voltage drop across the reactor will be very low and a large harmonic voltage will appear on the line side of the reactor. With a double circuit d.c. line, since the two outer conductors are identical, the analysis above can be made separately, and the results combined. This will not allow, however, - for the line connecting the earth electrode to the converter through which the unbalanced harmonics will flow [Figure 10.20 (c)]. An assessment of this part of the system can be made with the equivalent circuit shown in Figure 10.21 (d); the impedance of the electrode line is given by: Loe=Zee (10.58) where Z,, is the characteristic impedance, y, the propagation coefficient, and /, the length . tanh y, . I, We TOT e eee eee eaten eee Co 180 HARMONICS L, ansmission line of length l or equivalent angle ¢1=(o+jB)l Rectifier Loe link Distance x or angle x PINT ITHY = = = = — 4 ar eorany | Jovertor ly poet li] i] (&) Eguivalent circuit of dc. system for n& harmonic 27 = Z Sinhv2y v 2Y a | (pz = + Y fonh(Vzy¥/2) 2 (vZY/2) | 1 ero = Z fenh(Vzy/2) \ 2 W2¥/2) (¢) Equivalent T network ©) Line 1 = @) () Equivalent IT network | ERRATA: Figure 10.21 Under (5), for 2¥ read ZY Under (d), for LdR read Lyp Zen) Figure 10.21 (a) (6) (c) (d)_ Harmonics in d.c. transmission line ating \ I | i i | F cc 1 — DISTURBANCES CAUSED BY HARMONICS ON THE D.C. SIDE 18] Figure 10.22 Proposed low Pass filter to shunt the harmonics on d.c. side of the electrode line. The electrode line at the invertor end can thus be allowed for by adding the term Z,,), into the right-hand side of equation (10.48). The rectifier n harmonic current will then be given by V, Typ eo nn cece ccc eens (10.59) OF LoytZyeytjon - Lap and the current harmonic in the electrode line at a distance x from the electrode will be given by 5 re (10.60) Taya Cosh yf The current and longitudinal voltage harmonics along the main transmission line will be given by equations 10.53 and 10.54 or equations 10.56 and 10.57, respectively. The voltage harmonics in the electrode line may be neglected. 10.2.4 Disturbances caused by harmonics on the d.c. side, and methods of reducing them From the above discussion, the reactor alone will not be effective for suppression of the harmonics in overhead lines, and large harmonics in the line may well cause severe distur- bance in communication circuits. It should also be remembered that overhead lines cause much greater disturbance than cables. Thus in case of overhead lines it is necessary to provide a capacitor beyond the reactor, which will shunt the harmonic currents and effec- tively suppress the harmonics in the transmission system. According to the view of Russian investigators,422) the line reactor alone is satisfactory for cable systems but in order to limit surge currents, the inductance must be sufficiently high, c.g., not less than 1.5 to 2H. The value of inductance cannot be the same for all h.v.d.c. cable systems, however, since its value depends on the ratio between voltage and current and the amount of cable capacitance. Investigations by ASEA have shown that 1H is frequently suflicient; the value of inductance in the case of the Channel scheme is 0.95 H. For overhead lines, a smoothing device resembling a low-pass filter has been suggested, “22) * : ean ea i ea tt eterna 182 IIARMONICS as shown in Figure 10.22; the capacitor C, is large cnough to absorb harmonics and the reactor L, can be inexpensive. This type of filter can ensure high damping over the whole invertor audio-frequency band. The Soviet work suggests that for transmission lines, 600 to 1,000 km long, with a line voltage of 800 kV, a reactor of 1 to 1.5 H and capacitance C, up to 0.5.F, will suppress the harmonics in the overhead lines so that they do not exceed those of electrified railways. Harmonics can also be suppressed by providing separate tuned filter circuits for 6th, 12th... ete., harmonics, after the reactor. As with the case of harmonics in the a.c. side of the system, harmonics in the d.c. system will consist of balanced harmonics, which return through the other outer conductor, and unbalanced harmonics which return through the earth. Clearly, the electromagnetic induction in a communication circuit from the balanced harmonics will be subject to substantial cancellation since they are equal and opposite in the two conductors, whereas that from the unbalanced harmonics will be large since they are in phase in the two con- ductors. In a mid-point carthed system (Section 10.2.2) with two bridges displaced by 30°, a large proportion of the harmonics will be unbalanced and return through the earth. In the earth return system all the harmonics will be unbalanced. Since it is not economical to increase the phase operation to more than twelve (Section 10.1.4.1), clearly the best arrangement is to provide at least two bridges displaced by 30°, on each side of earth, so that both sides are in phase and each side has a twelve-phase operation. In this case the harmonics in the d.c. line will be those appropriate to twelve- - phase operation and will be balanced. In a mid-point earthed system, consideration must be given to emergency operation when only one side is working; although in this case the harmonics wil! be substantially reduced (since the voltage is reduced by half), all the harmonics will be unbalanced and will thus return through the earth connection. Furthermore, any unbalance to earth of the two sides will cause what otherwise would have been balanced harmonics to become unbalanced. In the case of a cable transmission system, the cable sheath and armouring will provide an efficient shield (Section 10.1.3.6), and in spite of the earth or sea return the disturbance will be small. Also, in the case of sea return the coefficient of mutual inductance for unbalanced harmonics will be small since it decreases with increase in conductivity of the return path and, in general, the ocean has a much larger conductivity than the earth. In the Gotland scheme,“19 calculations showed that the distance between the power cables and the telephone cable should not be less than 350 m, and a nearest distance of 2,000 m has been established. As is well known, the power cable operates at 100 kV, 200 A, with a sca return, and has twelve-phase operation. Disturbances in communication lines will also be caused by current harmonics in the lines from the converters to the clectrode. Ina mid-point earthed system, only the un- balanced harmonics, if any, will flow through the electrode line, whilst with earth return all the harmonics will take this route. A special case can arise when the main line is a cable and the electrode line is overhead. The harmonics may be too small to create any disturbance along the cable route, but may be large from the point of view of the overhead electrode line. aaa ee a _— a a ‘nema = . jaan Bonin Stn bai teste Fuk = SL ite DISTURBANCES CAUSED BY HARMONICS ON THE D.C. SIDE 183 This difliculty can be overcome by providing a capacitor connected between the electrode line (at its converter end) and earth. A large proportion of the harmonics will now return through the capacitor, provided that it is of sufficient size, rather than through the electrode line. If a capacitor of size C is provided, the harmonic current which will pass into the electrode line will be given by: 1 1 a ateteelle GoC Zant) 0 ; qe eee (10.61) where Z;,), is the electrode line impedance at the n™ harmonic frequency. Such a capacitor will also serve to protect the electrode line from atmospheric surges. The size of capacitor for the electrode line of the Gotland scheme“) is 100 F. Tove=T ey + — = = — ce CHAPTER I1- Insulators References : (109) (125) (170) to (185) inclusive, 11.1 General The development of insulators for extra-high alternating voltages has taken place over many years. This development has been based mainly upon experimental data because of the extremely variable and unpredictable nature of surrounding atmosphere and has led to an assessment of d.c. insulator performance by comparing it with corresponding perform- ance on a.c. under the same range of ambient conditions. Insulators at present in use are made of porcelain or of pre-stressed glass, the latter having certain advantages. The glass insulators shatter completely after rupture by elec- trical, thermal or mechanical disturbances, which makes for easy location, whereas porcelain may only develop a fine crack or chip and may present a continuous source of trouble which is difficult to locate. Also, the dielectric strength of glass is about three times that of porcelain. Figure 11.1 shows a number of different types of insulator in use for high-voltage trans- mission purposes (supplied by courtesy of Messrs. Taylor, Tunnicliff & Co. Ltd.). 11.2 Insulation level of a d.c. transmission line The insulation of a transmission line involves the following basic considerations ;(109 (i) Ability to withstand the Operating voltage under abnormal weather conditions. (ii) Ability to withstand transient over-voltages arising from faults and switching operations. (iii) The insulation should be such as to reduce the likelihood of outages due to lightning strokes to an acceptable figure. To minimise outages due to direct lightning strokes, an overhead earth wire is commonly used. The likelihood of indirect strokes increases if the tower footing resistance is high; this is dealt with by ensuring low tower footing resistance or even providing counterpoises in the form of conductors buried in the ground and connected to the towers. On high-" voltage lines these measures generally reduce the lightning outages sufficiently to make it unnecessary to increase the insulation level above that required from normal voltage considerations. In a d.c. system, over-voltage will occur during the various invertor, rectifier and line faults, converter starting and shutting down operations, etc. It is technically possible, ina d.c. system with good grid control arrangement, suitable size of reactors and filter circuits, to reduce the internal over-voltages to insignificant levels. Now an insulator can withstand short-time transient over-voltage of considerably higher magnitudes than the normal operating voltages and it is likely that when the internal over-voltages are not large, the ability of the d.c. line insulation to withstand the operating voltage under all climatic con- om | | | t INSULATORS - 185 Galvanised malleable cost iron cap loughened glass shed Cement joints a i | lo cia. Galvanized steel pees ; Figure 11.1 (@) Suspension or tension insulator aay Cement join Galvanized Toughened steel pin glass shed /1 dia. Figure 11.1 (6) Heavy duty tension insulator ~— = Sa an da acne nena aan Seana et as ANC cath SA in et ina 186 Galvanized. malleable iron Resilient pad cap Cement i 2 joints} 6 « Galvanized <° sleel pin 9 10 dia: nT Figure 11.1 (c) Antifog type suspension insulator ditions will be the main determining factor for insulation level. If this proves to be the case, internal over-voltage in other words, if the ratio, of a system is reduced below the ratio operation voltage transient flashover voltage — of the insulation, there will be no further saving in insulation continuous withstand voltage and any further expenses involved in reducing the internal over-voltage will have no justifi- cation. Thus an over-voltage factor of 1.7 has for a long time been considered as a practical minimum value and has also been fixed for Stalingrad-Donbass transmission system. 42 On the other hand, in view of the experience obtained on Gotland scheme, Engstrém of ASEA, Sweden, suggests that much lower values of over-voltage factor can be obtained without extra cost. With the ASEA system for example, the grid control will limit the rate of voltage rise, thus reducing starting over-voltage, which generally determine the factor of 1.7. This control will not result in any saving in the aerial insulation but would have a beneficial effect, for example, on fault clearing operations. 11.3. Insulator impedance The impedances of a chain of insulator discs consist of very high internal leakage resis- tance, surface leakage resistance, capacitance to carth of each disc, and capacitance between discs. The surface leakage resistance is an extremely variable quantity depending on surface | oma) cot Et ee ! INSULATORS 187 4 holes equally Machined surface oe ae 5 dia. ec./ Drainage slot 3!) > / M.C.L. INS > “\ Galvanized A if fp LoL : fi Y: Sh i t Aholes equally spaced 94 dia->| “MCI Galvanized Pee ee ste }-S aha PCAN FP Machined a 6F dio 4! surface Figure 11.1(¢) Post insulator MF ani st ade ornament en a athena, Lire 188 INSULATORS contamination and atmospheric conditions. In good weather, in the a.c. case, the voltage distribution is determined by the capacitance distribution and, as is well known, results in the greatest voltage appearing across the insulator disc nearest to the line. This is not necessarily the case under wet, contaminated conditions when the greatly decreased, and highly irregular, surface leakage resistance may well be the deciding factor in voltage dis- tribution. With d.c. operation of an insulator string, the surface insulation resistance is responsible for the voltage distribution as no capacitive current can flow under steady state. Under dry, clean conditions thé insulator leakage resistances are very high, and the voltage distribution might be determined by the direct electric field distribution. Swedish experi- ments recently have shown that the distortion of field is such as to cause a very substantial proportion of the total voltage to appear across the insulator nearest to the line: in the case quoted, about 40 per cent of the total voltage appeared across the line insulator of a six- insulator string. In both d.c. and a.c. cases, however, it is the performance under the worst likely conditions which determine the insulator rating. The factors which are thus paramount in applying insulators to d.c. transmission problems are the nature of contamination accumulating on the insulator surfaces in practice and its distribution, and the direct voltage breakdown phenomena experienced with insulator strings. : In a.c. practice there has been a tendency to eliminate grading rings as it was realised that leakage resistance was of paramount importance and reduced the effect of better capacitive grading. Under impulse conditions, however, the capacitance distribution may have some small influence even under wet, contaminated conditions; this is true for either a.c. or d.c. But at extra-high voltages, some form of shielding for the insulator adjacent to the line is considered necessary because of radio disturbance caused by corona across this insulator even in good weather; radio and television interference phenomena from high-voltage d.c. lines are at present under investigation. 11.4 Insulator contamination and wetting“70 G7» Contamination of insulators is naturally of most significance in the polluted atmosphere of heavy industrial arcas and in salty atmospheres near the sea coasts. Most forms of industrial contamination, such as soot and cement, are not very conductive so long as they are dry; they, together with salt, pass into solution when moistened by rain, mist or fog and then produce a conducting film; thus the worst kinds of contamination are those which contain a high proportion of soluble matter or are hygroscopic, such as magnesium chloride. Rainfall, if continuous, has the beneficial effect of washing the insulators, and thereby diluting the conducting film of material and producing relatively low conductivity; the Worst conditions are caused by mist, fog and light drizzling rain which produce no washing effect. A region with seasonal rainfall experiences poor conditions because of accumulation of contaminating material during the rainless periods. In many cases, soluble salts such as flyash, cement and phosphates accumulate with large quantities of wind-blown soil; this type of mixture bakes in the sun on the insulator surface and prevents any underlying soluble salts from being washed away. The materials with cementation propertics are the most'serious in this respect. : cnet ese iil la is Aas deni at atacand ee A” r [ [ voltage supey in igeaot d, and Jittons pping vf adtavn cdBsat adkve esome ° W be oride, Ac ishing fafon ospoh soil; metronome ntl An Sine Stes, esac DRS aaa ai ects at it a hae ne lade oe, DEVELOPMENT OF INSULATOR BREAKDOWN 189 11.5 Development of insulator breakdown®70 The results of much experimental work show variations in leakage current over insulator surfaces over a range of the order of 108. The variation is from the order of 0.1 wA under dry, clean conditions to several mA under dirty conditions in fog or mist. The distribution of contamination will clearly be irregular, and those areas of insulator surface with the lowest conductivity will be subject to the greatest voltage stresses. If part of the leakage current path becomes highohmic or breaks up, for example due to the drying action of the Icakage current, a very high voltage will be applicd over this gap. This voltage y may be found by where V and R are total string voltage and resistance, r the resistance of the highohmic gap. An arc with the current V/R will be caused and bridge the gap. If this current is large enough, the foot points of the are will be subject to a considerable drying action, causing the arc to move to fresh areas with surface conditions suitable to carry the foot point leakage current concentration. The wandering are will have a tendency to prolong and might even- tually lead to total breakdown. The arc may stop at metal armatures. But if the arc voltage drop is small, the rest of the string will be subjected to over-voltage, which might cause new arcs and eventually a cascade flash-over of the string. Such a discharge tends to originate on the underside of the line insulator, which is comparatively clean, and where the line clamp arrangements produce high leakage current densities. The breakdown conditions can be aggravated by the fact that the cleaner and drier parts carry greater voltage and dissipate greater energy, and thus tend to become even more clean and dry: the voltage across these areas thus tends to increase continuously until breakdown occurs. The sequence of events above is by no means certain, however. If sequential local breakdowns do not occur quickly, the first spark(s) will be extinguished as the increased current burns or dries up some of the contamination, resulting in an increase in the total resistance and consequent decrease in the leakage current. Local sparking has also been observed to occur for several minutes before developing into overall breakdown, but gener- ally the initial spark either spreads quickly into an arc or becomes extinguished, the latter case resulting in a current surge across the insulator surface. This isa particular characteris- tic of insulators and it has been observed with a.c. insulators,72) that the flashover of a chain or stack does not generally occur until these current surges attain a crest value of 100-150 mA: this has been found to be the case with d.c. insulators also. These phenomena lend themselves particularly to a.c. working since the current zero every half cycle in this case produces spark extinction and thus a chance of recovery. This may be the reason for the d.c. breakdown voltage of a contaminated insulator being lower than the crest value of the alternating breakdown voltage, as has been shown by various tests discussed in section 11.7. When an insulator chain is first put into service, it is clean and the conductor side insula- tor tends to carry the greater part of the total voltage. As the insulators become dirtier, the conductor side insulator tends to keep itself cleaner than the others: this arises because oO i 190 INSULATORS its higher voltage drop experimentally verified produces greater percentage losses. Thus, even in contaminated conditions, the greater part of the voltage is likely to appear across this particular insulator, an effect which is enhanced by the action of gravity in depositing material on the more exposed upper insulators, i.c., the carth side insulators of a chain, A further consequence of local sparking is surface heating. If the combined effect of heating and voltage gradient are sufficient, puncture of the insulator can occur because of differential heating and expansion, and increased local voltage stresses. Corrosion of the metallic parts of insulators and insulator strings is more likely with d.c. than with a.c. Corrosion is increased whenever intense corona and sparking occur on account of the production of ozone and acidic atmospheric pollution, especially when the metallic parts are shielded from air currents. Local destruction of galvanizing may occur because of sparking, with subsequent rapid corrosion of the underlying metallic surfaces. Insulators provided with hoods and extensive corrugations on the under surfaces are known as antifog insulators [Figure 11.1 (c)]; the corrugations provide additional leakage distances, and an area protected from rain and deposition which is capable of sustaining the necessary voltage until the exposed surfaces of the insulator have been washed to some degree by rainfall. The protected leakage surfaces (corrugations) must not have air-gaps which are too narrow between their various circumferences, since their value will be lost if local breakdown occurs. Another approach to insulator design which may be of significance to d.c. operation is to minimise the protected areas so that the insulator surface is subject to fairly uniform exposure to contamination, wetness and wind, and hence to uniform stresses; this approach can only be suitable for locations where there is a slow rate of deposition and regular rainfall, ie, where the insulator surface is not likely to be highly conductive. Clearly the corrugated type of insulator is most useful for highly polluted and coastal areas. 11.6 Causes of surface contamination and the effects of electric field on contamination Since contamination is the cause of low breakdown values for insulators, it is important to know ‘the extent to which direct voltage influences the contamination compared with alternating voltage under the same ambient conditions; the relative magnitudes of the forces exerted by the electric field of an insulator on dirt particles, compared with other natural forces, are also of importance. The forces exerted on dust and dirt particles in the air surrounding an insulator are of three kinds :473) (174) 7 (i) Gravitational force, which obviously causes deposition on the upper surfaces of the insulators. (ii) Wind pressure, as given by Stoke’s formula, Fy Xn. ry. where r is the particle radius, v is the wind velocity, 7 is the coefficient of internal friction of the air. (iii) Electrostatic forces. Charged particles will tend to move in the dircction of the field of the insulator towards the positive or negative pole, as appropriate to their sign. The magnitude of this “coulomb” force is, F,=e.n. § c where ¢ is the gradient or electric field strength in V/m. [ ; CAUSES OF SURFACE CONTAMINATION 19] Thus, ¢ is the charge of the electron (1.59.10"!? coulomb), t Bross nis the number of elementary electron charges carried by the particle. ‘Er The direction of this force coincides with the direction of the vector €, so that the hain. particle executes an oscillatory motion in an alternating field and a unidirectional eat of i one in a direct field. Actually under alternating field the oscillatory motion of the [ particles will be superimposed by a slow particle drift. This is due to the rectifying property of corona. Positive and negative corona currents may differ slightly in size ithd.c. and the difference causes the drift. t on i Force will also be exerted because of variations in the strength of the electric hal the field. All particles, whether charged or uncharged, whose permittivity differs from ¥ occur that of air, produce a distortion of electric field at their surfaces. In a uniform field the forces so produced are balanced, but if the field is non-uniform the forces are unbalanced and the particles tend to be drawn into the strongest portion of the field. Ina disc-type insulator, the ficld is strongest near to the link arrangements and, under ine.the both d.c. and a.c. conditions, there is likely to be more contamination at such points, [se due to this cause, than on other underside surfaces, assuming that no corona occurs. Witich The magnitude of this force is given by if local K-1 ifgee F,=27r3 & ‘Kar: grad ¢? Hi T WL to { For equal values of direct and alternating (r.m.s.) voltage conditions, there will be no fesses, difference in the value of F,. Thus the difference between the direct and alternating iE field conditions lies in the additional influence of “coulomb” force, F,, in the case i i of the direct field. ared The relative magnitudes of the forces above in influencing dirt and dust deposition are clearly of importance. A Russian estimate? of the different forces acting on particles of various diameters, has shown that wind pressure is the dominating factor; even in low winds, of the order of 3 metres/sec., the effect on particles of diameters greater than 1 micron ir exceeds by one to three orders of magnitude the effect of the electric field on highly charged particles. The approximate diameters of some of the common impurities suspended in the bs atmosphere are given in Table 4.73) Clearly, gravity plays an important part in the case of cy rural \ 4 TABLE 4 ibe Approximate sizes of different particles in the atmosphere ow the Nature of particles Diameter in microns [ Smoke .. oe 7 o oe *. oe 0.001-0.3 = ‘ Inorganic Fumes .. +h ale 24 ae ws HF - 0.01 -1.0 Dust... tL a tis |b ae i 1.0 -100.0 Bacteria 1.0 -10.0 Organic Plant spores. ae i - 10.0 -20.0 t Pollens .. +s + + on 15.0 -50.0 FH) |) esi EI) Me ae ee 1.0 -50.0 Water Mist sls i A Ae 7 a a 50.0 -100.0 Drizzle ae ae 7 ee le 100.0 -400.0 Rain 400.0 -4000.0 a t 192 INSULATORS Upper side of the insulator Underside of the insulator 500, A 4 - qr 09 400 ==88 | > 7 \e x S$ c oy < S $300 L S & as C 8 AH 30 AP &200 Ss ~ 3 8 tS 3 *| 8 Aries a. |8 Q /00 a] & Te = SS S oeeets eee ratty? 0 0 a4 leo fe 0-70 0 30 0 10 2 30 40 Applied vollage (kVmax/unit) Applied vollage (kVmax Junit) Figure 11.2 Influence of direct and alternating applied voltage (max.) on quantity of deposited salt (Japanese tests). Density of salt in air: 15 mg/m, Indoor at no wind. — Test sample: standard 254 mm suspension insulator . the larger particles. Thus for outdoor insulators, wind and gravity are the dominant factors, the effect of field being relatively insignificant in determining insulator pollution; for indoor insulators, electric field is of considerable importance. Corona discharge also plays an important part in the movement of particles.973) Corona discharge creates a large number of electrically charged particles which move away from the discharge points causing what has been called an “clectric wind” in the form of air jets which induce local air circulation. This electric wind is much stronger with direct than with alternating voltages, and is a substantial contributor to the uneven distribution of dust on insulator surfaces. ASEA have conducted experiments on the behaviour of particles in known electric fields,“\7) the main observations being as follows: (i) In a field without corona, particles are drawn in the direction of increasing voltage gradients (attraction). This is the case with both a.c. and d.c., of both polarities, The force of attraction appears to be proportional to the r.m.s. value of the voltage. (ii) Ifa particle touches part of the electric system, e.g., the insulator surface, it is repelled by d.c. of both polarities, but not by a.c. in which case it cleaves to the insulator surface. (iii) In the case of d.c., particles can be shown to be repelled even befe ore touching a stressed surface if the surrounding atmosphere is slightly conducting, e.g., in dense fog. (iv) Ina field with corona, particles are pushed in the direction of decreasing potential gradients (repulsion). This is the case with both a.c. and d.c. of both polarities, although the force is about ten times greater in the d.c. case. With d.c., the repul- sion effect can be observed not only on the clectrically live parts, but on the earthed sections as well, although to a lesser extent. This is one further effect giving rise to | | | | | | | aan) CAUSES OF SURFACE CONTAMINATION 193 differential pollution of an insulator string; the insulator sections which are in the vicinity of the live conductor tend to maintain themselves in a cleaner condition, whilst dirt particles pushed away from the live parts tend to accumulate on the sections nearer to earth potential, thereby becoming dirtier. The Electrotechnical Laboratory of the Ministry of International Trade and Industry of Japan,“ has published the results of experiments on insulators operating under widespread atmospheric conditions; some of these results are as follows. Figure 11.2 shows the results of measurements of salt contamination when a string of three standard suspension insulators (254 mm dia., 146 mm high) had alternating and direct voltages applied in a room con- taining a certain amount of salt in its atmosphere. The experiment shows that, under wind- less conditions, contamination of the earth side insulator is by far the greater when direct voltages are applied, than with alternating voltages. From this result it is certain that precipitation will take place upon support insulator bushings, etc., at d.c. indoor substations. With outdoor substations, the effect of wind has to be considered, and Figure 11.3 shows the results with a similar insulator string which was placed in a wind tunnel, with a wind velocity of 4 m/sec and in which the salt solution was sprayed by means of an atomizer. aan Around the normal working voltage (10 to 15 kV/unit), the contamination tends to be less a with d.c. than with a.c. The ratios of the amount of salt deposition on the lower, underside surface of insulators operating at normal working voltages as against the same surfaces of voltage-free insulators were 1.05 and 1.11, for d.c. and a.c., respectively. Corresponding ficld experiments, under wind conditions with a high salt content, showed that the salt deposition on the underside surfaces of the insulators of a three-insulator chain to which was ma = = — vail io Upper side = ; Under side A : — ae. —— dc. so—t----- ae. tt 4AQB RTT ac. {i > Cc 7 XS 40 wr a § S id: S 30 x LB a : ~~ 20 SS oF 8 & Cy} . & 10 Ys bao Q 7 8 & oO oo 0 9030 0 0 2 30 Applied voltage (kKVmax.funit) Applied voltage (kVmoxfunit) Figure 11.3. Influence of direct and alternating voltage on quantity of deposited salt (Japanese tests). Density of salt: 2.0 mg/m?. Wind velocity: 4 m/se¢ | | | ss pibtasiaibas iit en nS nn as it la 194 " INSULATORS ai ‘N s ‘N N /O N : N NY TN a 7 © LASERS é 7 : Figure 11.4 Distribution of A & { \ salt deposit on the underside NG 7 4 of an insulator under the in- ~ o "] fluence of direct and alterna- a 5 ting voltage (Japanese tests) & | ‘ vU 1 et Xn---- Mw AC, oO o---—-de. | ws) . No voltage || (CCEA Tae B C- D Fort of insulator surface applied 45 kV direct voltage and 45 kV peak alternating voltage Were, compared with voltage- free insulators, in the ratios of 1.03-1.11 and 1.23+1.37, respectively. Figure 11.4 shows the distribution of salt on the underside surface of the middle insulators of the chains used in the experiment immediately above (45 kV max/chain); the deposit is concentrated around the pin in the case of a.c., whilst it is more uniformly distributed in the case of d.c. Under rain test, it has been observed that direct field, independent of polarity, tend to counteract the wetting since raindrops impinging on the insulators become charged and are consequently repelled by “coulomb” forces. As with dust particles, they are even subject to repulsion before touching the insulator surfaces, It may generally be concluded from the results of the tests above, that the contamination of outdoor insulators is not likely to be any more severe under direct voltage than under alternating voltage conditions in practice. a vv eos > to ny | | FLASHOVER AND WITHSTAND VOLTAGES OF INSULATORS 195 11.7 Flashover and withstand voltages of insulators According to investigations in the U.S.S.R.,47? dic. line transient over-voltages can be reduced to 1.7—2 times the operating voltage, and in practice most of these transients take the form of a rapidly rising voltage which does not change sign and has a lifetime of the order of 0.1 sec. Experiments on a string of five, suspension-type insulators (type PTs-7) and on a single insulator with voltages similar in form and duration to the surges occur- ring on a d.c. transmission system, showed that ihe wet flashover voltage level under prac- tical conditions was unchanged when the laboratory test conditions produced a voltage rising at a rate of 20 per cent/sec. Figure 11.5 shows the change in wet flashover voltage when the rate of rise of applied test voltage is altered: taking the value of flashover vol- tage at 20 per cent/sec rise as 100 per cent, it can be seen that the flashover voltage reduces by about 15 per cent at 4 per cent voltage rise/sec, and increases by about 10 per cent at 50 per cent voltage rise/sec. The same effect, although to differing degrees, will apply to other types of insulators and insulator arrangements. There are two main conditions under which surges can be impressed on the insulation of the system: : (i) When the system is energised and the insulation has already been subjected to operat- ing voltage stresses, the surge over-voltage is then applied. (ii) Application of the over-voltages when the line is not energised. The relative severity of these two conditions depends on the weather, insulator design, and contamination of the insulator. Generally (i) represents the less severe conditions because the normal operating voltage conditions would have heated and dried the insulator particularly the corrugations on the underside, and would have decreased the insulator conductivity. On the other hand, should local discharges be occurring in the case of normal voltage energisation, the insulation would be very susceptible to complete breakdown on application of an over-voltage surge. The difference in wet flashover voltage, with and without previously applied operating voltage, will generally be small (less than 10 per cent),“77 but should nevertheless be taken into consideration when testing. 8. et le _ 7 Sr00% — 334377 oe 7 © ofp aaer & 8041+ Spat 8 L « a & 604 ROO 4 8 12 6 90 2 9% 32 36 40 44 % voltage rise per second Figure 11.5 Relation between wet flashover voltage and rate of rise of voltage (string with cight insulators of type PTs-7. Negative polarity. Rain intensity: 3 mm/min. (Russian tests) : f' uns onan nals leet Nn a 196 INSULATORS TABLE 5 Ratio of flashover voitage to maximum withstand voltage for different types of insulators (Russian tests) Supporting rod Supporting types with Cap and pin type, with less developed surfaces Insulator type suspension developed surface |———_—_ —— (PTs-7) (KO-400) (Sh D-35) (ShT-35) Polarity Bal ae +. - + - + - + > Ratio of — flashover voltage at 20 per cent rise/sec to maximum withstand voltage .. Table 5 shows?) the ratio of flashover voltage (rising at a rate of 20 per cent/sec) to the maximum withstand voltage as obtained from tests on a number of insulators under an intensity of rain of 3 mm/min. The withstand voltage is defined as the maximum voltage which an insulator can withstand for thirty minutes without flashover. During these tests, the voltage was first adjusted and rain was then directed on to the insulator. It can be seen from the table, that the ratio is high for insulators with only a slightly developed sur- face (type KO-400) compared with those with a highly developed surface (type IShD-35). Since the ratio is 1.7 (case of negative polarity) for suspension type insulators, as discussed above in section 11.2, it would be of no importance from the point of view of saving insula- tion of the line to reduce the internal over-voltages to less than 1.7 times the operating voltage. Table 6 shows the results for the variation in flashover voltage with change in the intensity of incident rain; the flashover voltage at a rain intensity of 3 mm/min has been taken as unity. The flashover voltage rises with decrease in the rain intensity and vice versa, but the effect is less pronounced for increased intensity of rain and is more pronounced on insulators with less developed surfaces. Figure 11.6 shows the flashover voltage of the suspension type insulators (type PTs-7), at a rain intensity of 3 mm/min, with change in the conductivity of water, the value at a TABLE 6 Flashover voltages at different rain intensities (Russian tests) Supporting rod Supporting type Cap and pin type with less with developed Insulator type suspension developed surface surface (PTs-7) (KO-400) (ShT-35) Polarity 1h ae tle + _ ++ | _ he | _- Ratio of flashover voltage At raintintensity of 5 mny/min at rain intensity of 5 and 1 — — mnymir to the flashover 0.9 | 0.97 | 0.98 | 0.94 | 0.98 | 0.97 voltage at 3 mm/min rain |—— — | intensity At rain intensity of 1 mm/min i, 1.13 | 1.07 1,23 | 1.19 | 1.08 1.06 Sn eeeeeeee FLASHOVER AND WITHSTAND VOLTAGES OF INSULATORS 197 WN Ie /-0 -—— | : °o ee °| 40 60 80 100 1290 140 (micro-ohm. cm)~! Figure 11.6 Relation between relative flashover voltage and conductivity of water (string of eight elements of insulators type PTs-7). Rain intensity (3 mm/min); x positive polarity; o negative polarity; A =50 c/s (Abscissa micro hm cm)—! (Russian tests) conductivity of 100 microhm-cm-! being taken as unity; the results are very similar irre- spective of the polarity of the d.c., and as between the d.c. and a.c. (r.m.s.) values. Table 7 shows results obtained at the H.V.D.C. Institute of the U.S.S.R. to compare De. Ratio of Di. Type Wel flashover | wet flashover Insclat average gradient | to 50 ¢/s. noms. eur RV per metre ¥ |wel flashover Dise-lype (\ stspension | ao bo rtd Rod- a lie 8% os Busbar 2 : txsulator 220 £0 - tt bt? # Flashover value Jor insulator 1 melre long, q Ue fen siyle the d.c. wet flashover voltages of a number of insulators with those for 50 c/s a.c. The values obtained in the two cases were very similar. It was also observed during these experiments that when the direct voltage was applied for a short period of time, the flashover voltage under dry conditions coincided almost exactly, for all types of porcelain insulators, with the peak value of the 50 c/s flashover voltage. When the appli- cation of voltage was of longer duration, i.e., exceeding 5 sec, the flashover voltages of some types of insulator were reduced by approximately 25 per cent. Fig. 11.7_ Relative alternating and direct wet flashover voltage of different types of i nsulators (from Russian tests) 4 5% Boe i i a, i ee ae 198 INSULATORS Fog tests have been conducted in the Swedish ASEA laboratories,475) with dimensions of 4 m by 4 m by 6 m was used in which fog was produced atomizer and fan. Pure water, and pure water to which had been added 4 of sodium chloride/litre, were used for the experiments. of 5 per cent/sce from zero to the desired voltage level, w for one minute. Table 8 shows the results of both fog obtained by first submitting the insulator to both rain or fog, and only after stationary conditions have been obtained, to voltage. The results thus correspond to voltage being applied to a line which has been disconnected for some time, and may well be lower by about 20 per cent than when rain or fog are applied subsequent to voltage energisation, which corresponds to the normal service conditions, Under fog conditions, it can thus be seen that the direct voltage withstand values coincide with the alternating voltage r.m.s, values, but under rain conditions the direct withstand values are equal to the alternating voltage crest valucs. “Langstab” insulators, with relatively small Screens, give direct with- stand values only slightly above th M.S. > one test, with 0.4 per cent salt fog, using 5 and 10 element insulai i i pected to rise linearly with the nu Japanese results(170) to the alternating y A fog chamber by means of an 0, 4 or 0.4 gms The voltage was increased at a rate here the voltage was kept constant and rain tests; the values have been mber of elements in the chain. also show that for clean insulators, the dir oltage crest value; for foggy and polluted conditions, voltage flashover values are approximately equal to the a.c. From the very large amount of evidence available, it seems leakage surfaces are more favourable for direct voltage. One further general point which may well be of importance is the effect of altitude; the higher the altitude, the lower will be the flashover Voltage, and in a.c. practice an allowance of 3,5 per cent reduction per 1,000 ft in excess of 3,300 ft is usually made. , however, the direct rm.s. flashover voltage values. that those insulators with large 11.8 Radio interference from insulators As transmission voltages continu tion. Such interference uneven voltage distributi € to be raised, radio interference demands greater atten- arises partly from the insulators due to sparkings caused by the on and partly from the line conductor. The frequencies involved TABLE 8 a.c. and d.c, withstand voltages under rain and different densities of salt fog (Swedish tests) od Fat Pipe water Salt fog Salt fog Salt fog planiard Con fog 0.04% salt | 0.4% salt 4% Sait Tie =6,3002em | =4002em | =1008em Insulator type —dec.}-+defac| +d, ac | +de.! ac | +d} ac. de. kV | kV [kv kv kV | kv kV K Cap and pin type i Selements .1 270 | 280 | 190 250 | 250 150 150 90 90 60 70 Cap and pin type 10 elements -- Langstab 75/14 300 | 310 290| i [ STABILISED INSULATORS ~ 199 range from 150 ke/s to 150 Mc/s and are a serious embarrassment to radio and television ' reception in the neighbourhood of the lines. 5 “No great amount of data is available so far, but it has been suggested“7) that if the insula- tion level of d.c. lines is determined by service voltage and over-voltage considerations, then ute 2 no higher radio interference will occur than for corresponding a.c. lines. i 11.9 Stabilised and other special types ef insulators“178) (179) (so) [ The difficulty of uneven voltage distribution caused by pollution and subsequent local of, sparking, may be overcome if each insulator disc or section is provided-with a suitable fixed be resistance in parallel. One of the most attractive ways of doing this is to apply a thin rf glaze of a semi-conducting metallic oxide in place of the ordinary glaze on the porcelain. in Insulators of this type have been developed and are known as “stabilised” insulators. The the stabilising effect of such insulating glaze is enhanced by its inverse characteristic of resistance against voltage; thus when a high voltage drop appears across a high-resistance patch on an insulator surface, the semi-conductor resistance tends to decrease thereby decreasing the likelihood of sparking. When applied, semi-conducting glaze practically eliminates local sparking and local radio interference, but widespread adoption is unlikely at present because of a number of undesirable properties and characteristics. Semi-conducting glazes tend to ' deteriorate gradually when exposed to the electrolytic action of most atmospheres, and are thus unsatisfactory at present for overhead lines. The glazes also have great temperature dependence and resistance decreases rapidly as temperature is increased; stability is thus difficult to achieve since an increase in current across a glazed surface will produce an increase in temperature which will tend to decrease the resistance further and thus increase the current. If stability is not attained by heat dissipation, and the resistance increased due to a decreased voltage across the affected section of insulator surface, arcing will take place with the consequent likelihood of fracture of the porcelain. Aside from these non-linear characteristics, a correct value of resistance for a glazed section is difficult to obtain because of the uneven nature of the pollution over the insulator surfaces and variation in the insulator surfaces themselves, especially in the cap and pin type; semi-conducting glazes have been most successfully applied with post insulators which have a more uniform surface area and sheds which are not too deep. The possibility exists of protecting a semi-conducting glaze by an insulating glaze; this, if successful, would solve the problem of deterioration. A further way of obtaining better voltage distribution is to make the body of the insulator of a conducting material,“7 or incorporate resistance in the insulator units in some other way. A truly satisfactory material for a conducting insulator has yet to be developed, and - constructional difficulties are very apparent with the second method, however. Porcelain and glags are both susceptible to the effect of moisture and have inherently poor surface characteristics. A number of other materials are better in this respect; polythene, for example, maintains a high surface resistance under humid conditions, but tests at Croydon (England)"7 have shown unfortunately that polythene insulators suffer deterior- ation under outdoor weather conditions. Further possibilitics which are at present being investigated are insulators based on polyester, epoxy or silicone resins, all of which are Strong mechanically. Sas £ bak a Cre CR eS ST eS eS a CHAPTER 12 Corona References: (83) to (86) inclusive, (102) and (209) to (215) inclusive Corona is the result of voltage stresses produced in the air surrounding a charged con- ductor; the voltage gradient is a maximum at the surface of the conductor and decreases logarithmically with the distance from the conductor. As the conductor voltage is in- creased, the gradient also increases until a point is reached when the air immediately surrounding the conductor breaks down and becomes an ionised conducting medium. Corona is known to be initiated at the positive electrode by incoming natural electrons driven at ionising velocity by the corresponding electric stress, and at the negative electrode by outgoing electrons liberated by the impact of incoming positive ions. Corona is an extremely variable phenomenon; apart from a dependence on factors such as conductor diameter, distance between conductors, type of conductors (solid, hollow, stranded, or bundled) which are constant for a certain line, it depends also upon the weather (pressure, temperature and humidity), atmospheric pollution and the nature of the conductor surface, e.g., roughness. The effect of rain, fog, snow, wind, etc., is so great that corona losses in bad weather can increase by as much as ten times compared with good weather conditions. 12.1 Critical corona voltage and voltage stress The voltage stress © ,,,. kV/cm at the surface of a smooth conductor of radius r cm is given by: (i) For a single conductor at a height of 4 cm from earth, Vv r\2 Vv § max=——— = -( 14+) w~——. ............... ; ind 2h ( +z) 2h Nae r. loge a Fis loge- where V kV is the potential of the conductor to earth. (ii) For two conductors at a distance of Scm between centres, and at a height of Acm from earth, max 7 h artes Tee} ! roel? TE} V V 4 : where + 3k and ~z kV are the conductor potentials with respect to earth. it insesasen, meer 2 u i a n wis. nm o id = ag? Co Es & I -& a ~ Nted i ' t \ an need roi amumaih scinsndinaanAiiaen allan Settle siete ist inicn CRITICAL CORONA VOLTAGE AND VOLTAGE STRESS 201 Corona commences when the voltage V reaches the critical value V,, so that the maximum gradient € ,,,, attains the critical value € , at which the air breaks down. It is clear from the equations above that the greater the value of r, S and h, the greater will be the value of V, for a given value of € .. Clearly the value of &, will not be constant but will depend upon atmospheric con- ditions; it decreases with decrease in pressure and increase in temperature, and vice versa. The breakdown stress also depends upon the radius of the conductor and decreases with increase in conductor size. Peak®09) established a density factor § to take into account the atmospheric conditions and also suggested that, for the same atmospheric conditions, the breakdown stress for visual corona is such that the stress is constant at a distance a from the conductor, where, so that, § =m. ©, [840.301 V(8/r)]..-...ceceeee cece eee (12.4) where € , is constant (=30 kV/cm) and is the breakdown value of air at a temperature of 25° Cand a pressure of 76 cm of mercury; 8 is the density factor which takes into account the deviation of atmospheric temperature T (in °C) and pressure p (in cm of mercury) from 25° C and 76 cm, respectively, at which values §=1. 3.92p Fas HI (12.5) Equation (12.4) when arranged in general form becomes, © HA. BAB. VA (8/d) oo cece ees (12.6) where A and B are constants, d is the diameter of the conductor. Various investigations have been carried out to determine the value of the constants A and B, which differ slightly from each other, and also differ for negative and positive polarity. © .=39.85-+-10.36 . /(3/d) for +ve corona} Wriitchead and Leet210, . (12.7) & -=40.35-+411.91 . (S/d) .. for —ve corona; & .=33.78411.5 .V(8/d) .. for +vecorona|, vio) €.=31.08+13.5 .V(8/d) .. for —ve corona, BYOWN veer teee ooo (12.8) =35.08+411.4 ./(8/d) .. for +ve ea (1) c ir 2 €.=31.68411.9 . V(3/d) .. for —ve corona Farwell?" ..-----++--- (12.9) The decrease in critical voltage stress with increase in size of the conductor does not mean that the breakdown voltage V, decreases; substituting the value of € , in equations (12.1) and (12.2) will show that V, increases with increase in conductor diameter. As corona appears, the corona sheath increases the virtual diameter of the conductor; the virtual diameter increases with increase in ‘corona so that the maximum stress on the air surrounding the corona decreases. No further increase in corona occurs when the virtual diameter is such that the maximum stress in the air surrounding the corona is just less than that required for ionisation of air. To increase the corona initiation voltage, hollow conductors, conductors with fillers, or bundle conductors may be provided. Tikhodeev@!2) has given a number of equations for obtaining the corona starting stress and voltage for smooth, spiral and bundle conductors: = =m t | f t E aiscciiuatasch mo : : 1 ead 202 CORONA (i) For a smooth, single conductor of radius r, the critical stress € ¢ for initiating corona on the surface of the conductor is given by: E2—2. E, loge Foal 4-— eee eee (12.10) € where E,= a , and & ,=22.8 kV/cm, k P,==760 mm of mercury, p=atmospheric pressure and A=loge (+2) 0.168. © 42 where Q denotes the emission of secondary electrons per ion and has a value of 10~, approximately. (ii) For a bundle conductor system, consisting of 7 smooth constituent conductors, the critical stress € ., is given by, Eoy=Eg— Ny voce ees ee leek Ut (12.11) at i & cn where E,,= —", & k & ,=22.8 kV/cm and £, corresponds to a smooth single conductor of radius r equal to the radius of One constituent conductor of the bundle under consideration {equation (12.10)]. A,==(a—1). sin vt Eat sept EHD loge £.| ; (E,—loge E,—1) where d is the distance between the two adjacent constituent conductors. (iii) For'a spirally wound conductor, the critical voltage & ,, is given by € ym 17 aol fel ae tld bel ddb lel dea kod (12.12) e where ¢ . is the corona starting stress of a smooth conductor of the same size. The above equation suggests that within the practical range of conductor sizes and the number of component strands, the corona starting stress of a spirally wound conductor is independent of the number of strands. Also, with the spirally wound conductor, the maximum field strength at its surface is 1 — times greater than that at the surface of a corresponding smooth conductor for the same ms, applied voltage. The value of mz varies within a narrow range of 0.69 to 0.71, increasing almost linearly with increase in the number of component strands from 15 to 30, and can be taken as0.7 for practical purposes. Thus the final coefficient of irregularity introduced by the provision of a spiral instead of a smooth conductor is given by, m=ny . m21.17X0.7=0.82 7 and is almost identical for all spirally wound conductors within a practical range. \ j GENERAL BEHAVIOUR OF CORONA DISCHARGE 203 ony The critical line voltage at which corona is initiated for different system arrangements is: (i) A single-pole line at a voltage V, kV to earth. 2h Vo=&o.r.m.logze-- Peet seen (12.14) _ where € . is given by equation (12.10) and m=nym =the coefficient of irregularity, and is equal to 1.0 for smooth con- ductors and 0.82 [equation (12.13)] for spirally-wound conductors. (ii) A double-pole line with poles at +V,/2 kV to earth. Ss 1 Vi=28..r.m.log.-- ; "EQ To calculate the critical voltage of single-pole and double-pole lines with bundle (smooth or spiral) conductors, Tikhodeev®12) has given equations for a “coefficient of bundling efficiency” given by: tahsedaeyl (12.15) leas: | Soa iy .. (12.16) where V, corresponds to a line made up of separate conductors of radius r and is given by equation (12.14) or equation (12.15), and V., corresponds to the same line with bundle conductors consisting of 2 constituent conductors each of radius r. Equations (12.17) to (12.20), inclusive, in Table 9 show “coefficient of bundling efficiency” for single-pole lines and equations (12.21) to (12.24), inclusive in Table 10, show the same coefficient for double-pole lines. The coefficients above relate to so-called “general” corona. When calculating the values for “local” corona, it is necessary to supplement the equations with connecting factors obtained from experimental results. a {| xl 12.2. General behaviour of corona discharge for positive and negative polarities In the equations (10.10) (10.11) and (10.12), no account has been taken of the difference in the critical corona voltages for positive and negative polarities. In laboratory experi- ments it has been observed that, for the same pressure, temperature and other ambient conditions, positive corona appears at a lower voltage stress than negative corona for small diameters; when larger diameters (the sizes used in high voltage transmission lines) are used, negative corona may appear at a lower voltage stress than positive corona. But under practical conditions, with wide variations in atmospheric conditions, pollution, etc., corona for either polarity may appear first. A significant feature of the phenomena is that the positive critical voltage is not much affected by weather variations, whereas considerable variation occurs in the value of negative critical voltage. There are differences in the appearance of -positive and negative corona. Positive ing corona appears as a bluish-purple glow spread uniformly around the conductor and pro- duces an audible hissing sound; negative corona appears in the form of a number of con- centrated discharges, generally termed as “brushes”, of reddish colour, brighter than positive corona and making considerably more noise. This difference is also found in a.c. corona when positive and negative corona discharges are scen separately by means of a stroboscope. Negative corona is very sensitive to the state of the conductor surface and the discharge a i (a Sam te a a a a lea #4 204 CORONA Table 9 Equations of bundling efficiency coefficient for different types of bundling arrangements of single pole line. Single pole line with bundle Bundling efficiency coefficient conductors Kn=Ven|Ve i ! + _V 27 mf L (2h - { log —- F d Kea rd 7 8 co : (12.17) j obo, | | t ath 7 log a —2 (log 3)" [toe >] fa h K,= =x 58, (12.18) [(i+3 5) tog 4 —2 Flog 2] : i_V ; 77-20 ; (2h? OP. toe ae h Kyw= ———“___x Sa (12.19) r h € : (1+2v35) tog 4 _V 2° OO F of log 24" Dra ae h K=-——? _, Se (12.20) r\, dh Se" (1+3v2 5) = concentrates upon the dirty and unsmooth parts of the conductor; also the amount of noise is considerably reduced as the quantity of dirt increases. The “brushes” occurring in negative corona are fairly evenly spaced along the conductor and tend to move back and forth along the conductor. As the negative voltage is increased the number of brushes is increased. When the humidity increases, the negative discharge ea [ [ indling of ductor ‘fsa rge far (rr pv athe oan : 2 = : Pal toute ated Att pass i ei Nat i hanes GENERAL BEHAVIOUR OF CORONA DISCHARGE 205 Table 10 Equations of bundling efficiency coefficient for different types of bundling arrangements of double pole line. Double pole line with bundle Bundling efficiency coefficient ‘ conductors Kn=Ven|[Ve +Wo ~Va 4 dt lo, = . os : FO|O O15 Ko= mle ey «£2. 221) A (1425) log 5s Sie Hi Lome! ° ; VII ITIADT AIT IT? 1 | ite silt Lees / Lar a O07 log [vw 5 »|-2 ea Jerre re aes “6 w(t Ky= : ever ee x 52.12.22) (1+3 5) log 4 -2 log 2. log {5-1 _} ; ee eee cee £8. ana se : . Ss (142v35 7) bs; TOF SF i } log { st, 1 } } / By 2h 2 Ky= V2rd* | [1+ (S/2h)"} x Eas (12.24) (143v2 4) tog aa loses its appearance of regular “brushes” and there is a tendency towards a continuous glow effect. It has also been observed that corona bows the negative conductor towards the positive conductors, which in its turn bows away from the negative conductor as though wind was blowing across the pair of them. If the conductors are slack, the positive con- ductor has a tendency to vibrate in a circular path. i | Pp 206 difon dW) 10 Fn — rm. o_o ~— 02 01 }- o4 ‘ os os} 0-06 006 0-08 og 002 t | 202) cof o-0 = @ 50 00 50200 250 500kV max. 060-100-180 P00 250° 500 max QAL (OY DE, Fuit-tine curves: positive polarity. Dash-line curves: negative polarity, Figure 12.1 (a) and (6) a.c. and dc. corona losses on a 27.7 mm diameter conductor, shown as a function of the peak voltage to earth. The figures adjacent to thi oO : © curves relate to different weather conditions indicated in Table 11 (Swedish tests)(85) 12.3 Experimental results of d.c. and a.c. Recent experiments have been carried out in Sweden, France and Russia to determine initiation voltages for d.c. corona, and the accompanying losses, with increasing voltages. ‘ Since extensive data is available for the effects of a.c. corona, these observations have mainly been made to form the basis of comparison between d.c. and a.c. corona under similar atmospheric and conductor surface conditions. In Sweden®5)-the observations were made on a line 480 m long with space for three conductors. The conductors were of stecl-cored-aluminium with diameters of 27.7 and | . . . . ae 33.9 mm, and included both new conductors and ones which had been in service for a few years. The measurements were taken with a voltage applied between one conductor and [ corona compared earth, and with the other two earthed. Figure 12.1 (a) and (6) show the a.c. and dc. loss measurements for a conductor with a diameter of 27.7 mm as functions of peak voltages for different weather conditions in accordance with Table 11. It can be seen that the corona starts’at approximately the same voltage for both direct and alternating (peak) voltages. The losses with a.c, inercase much more rapidly than with d.c., however; the losses and critical Voltage for positive polarity show approximately the same dependence on 4 EXPERIMENTAL RESULTS OF D.C. AND A.C. CORONA 207 weather conditions as for alternating voltage, whilst for negative polarity they are more affected by weather. During fine weather, the losses are lower for negative polarity than for positive polarity and in bad weather the conditions are reversed. French® investigations were carried out on two parallel conductors of 16 mm diameter (147 mm? cross-section), 100 metres long and with 4 metres spacing. Conductor A had been energised for several months about three years before the tests began, and before each direct current test it was energised for 24 hours so as to obtain more stable results for the losses. Conductor B had never been energised with a.c. and was used only for direct current tests; it had, therefore, not been submitted to any ageing process. Figure 12.2 (a) and (6) show the loss currents J, and J, in conductors A and B, in relation to +V, and —V,. Figures 12.3 (a) and (5) give the a.c. and d.c. losses under similar conditions, one conductor being earthed and the other energised with direct voltages equal to a.c. peak paw shes Me SEN fete = - een rete eemnennnnmn sence een ie LET LE Ana ah gala tte i (6) Corona Loss Current I, in relation to Us and Us Fine weather : Mercury p=760-mm I, inpA h Temperature t= 30° C (a) In inp A. Corona Loss Current Is in relation to Ux and Up i Fine weather Mercury p=760 mm Temperature t= 30° C 500 . Up: 180 kV 400 X 400 I | f ! 300 11300 in Us| A160 ui ~160 kV an nes. ' L140 ye 200 71291200 if S| a t |_| _11c0 10 r_[o 5 Er iw ) Pool) eT, Sy 0 0 ° +50 +100" +150 +200 0 +50 +100 +150 +200 Figure 12.2 (a) (b) Corona loss currents in two conductors A and B, 16 mm in diameter, 100 metres long and four metres apart; both conductors energised (French tests)(84) tien Paha tenant ish iedin cae ame cation a cee i t i 208 CORONA i 7 in kV 1000 Ua_in 1000 1000, i ' C iM 0 be s 3 3 a 100 100 3 100}— Ss 7 7 = yn ® é 4 0 == % u ° ° ~ 10 10 ~ 10} 1 1 Crest Voltage 1 Crest Voltage), 100 150 200 in kV 100 150 200 inky (a) (6) Figure 12.3 (a) (b)_ a.c. and d.c. corona losses in two conductors A and B, 16 mm in diameter, 100 metres long and four metres apart; one conductor energised and other earthed (French tests)(84) voltages. It can be scen that losses with a.c. increase very rapidly as the voltage is increased and are considerably higher than in either the case of positive or negative polarity with dc. ; It is important to compare the d.c. losses shown in Figures 12.2 (a) and (6), which correspond to a case of a double-pole d.c. line, with the d.c. losses shown in Figures 12.3 (a) and (4), which correspond to the case of a single-pole line. Taking 160 kV as the voltage - for comparison, Figures 12.2 (a) and (b) give loss currents of Z, and J, equal to 195 pA and 210 A, respectively, for voltages of +160 kV on conductor A and —160 kV on conductor B. These currents correspond to the losses of 31.2 Win A and 33.6 Win B. Against this, the value given by Figure 12.3 (a) for +160 kV on A (B earthed) gives a loss of about 13 W on the value given by Figure 12.3 (b) for —160 kV on B (A earthed) gives a loss of about 6W. This clearly indicates that the losses per conductor for a double-pole line are con- siderably higher than the losses with a single-pole line. This is caused by a large current flowing between the conductors in the double-pole case. a | 3 L u a aa EXPERIMENTAL RESULTS OF D.c. AND A.C. CORONA 209 This last observation was demonstrated by extensive Russian tests(100) (213) (1) on experimental line Spans with single as well as bundle conductors as single-pole lines, with different distances between the constituen pole lines, and with different distances between two poles of a tests were carried out up to voltages of 700 kV between poles. The following are the over: by the fact that during the worst weather (which for corona implies and wet snow), the corona losses on d.c. lines rise by no more than i for a.c. lines they may increase by 100 times. losses on single-pole lines vary greatly if a gi for surge protection on the line, If there is a a.c. line for the same transmitted power, since energy transmission over two conductors requires a higher voltage; the latter fact results in much higher potential gradient on the d.c. line. Conclusion drawn in point (v) (c) seems rather doubtful, especially as the basis of comparison is not mentioned. A more realistic basis of comparison will be to consider equal line to earth voltage. Ifthe AC Power=3£Ep : J, where Ep and Tare phase values, the corresponding d.c. line of equal losses and equal line to earth voltage should have a power Pyc=2. Ep. (1.59), i.e, the third a.c. conductor is “split up” and added to the two others, Each d.c. conductor carries 1.57 and has a cross-section area of 1.5 times that of a.c. con- ductor. This line should, according to point (Vv) (2) above and because of its larger dia- Meter, have considerably lower corona losses than the corresponding a.c: line and sti’ be cheaper (50 per cent few insulators, simpler towers), 210 CORONA 12.4 Importance of corona losses in a.c. and d.c. transmission» In order to eliminate the need for employing bundle conductors or hollow conductors, it is necessary to revise the conductor cross-section above the most economic size with respect to the transmission capacity; this will not result in any higher transmitting capacity with a.c., since the capacity in this case is determined by the reactance even though, from economic point of view, a decrease in losses is obtained. With d.c., on the other hand, a corresponding increase in transmission capacity will arise: conversely where the trans- mission of a given power gives am economic cross-section smaller than that determined by corona considerations, it is also possible with d.c. to reduce the voltage whilst at the same time increasing the cross-section, whereas with a.c., the voltage is fixed by the requirement that the line shall be capable of transmitting the given output with an adequate margin of stability. For this reason there will be less deviation from the economic optimum with d.c, than with a.c, 10024 in 100 50 60 40 os 06 04 02 Of oo 60 100 50 200 250 200 max. “0 50 100 150 200 250 = 300kVinax. (2) AC (D.C. Full-tine curves: Positive polarity , Dash-line curves: Negative polarity. Figures 12.4 @ (6) Radio interference levels for a 27.2 mm diameter conductor, shown as a function of the peak alternating and direct voltage to earth. The figures adjacent to the curves relate to different weather conditions indicated in Table 11 (Swedish tests)(85) at Pe RE EOE 1 penne se ctpes, Bih racity ffn RADIO INTERFERENCE 211 12.5 Radio interference Figures 12.4 (a) and (5) show the results of Swedish tests‘6® on radio interference levels for the conductor of 27.7 mm diameter (the same as used for corona losses shown in Figures 12.1 (a) and (6), corresponding to weather conditions as shown in Table 11. It can be seen from the figure that a striking difference exists between the two polarities; with direct current with positive polarity the interference levels are of the same order of magnitude as for alternating voltages (peak values), but with negative polarity they are very much smaller. There are grounds for believing that the two curves for positive polarity which lie appreci- ably above the others should have less importance attached to them. Other investigations show®® that with a smooth polished, positive d.c. conductor, there is negligible interference, whilst with a negative conductor it is more severe (though less than in the a.c. case). Weathering and contamination increase radio noise from a positive . conductor, but reduce it for the case of negative polarity. Water on the positive conductor causes severe interference, but very little effect where the negative conductor is concerned. TABLE 11 Asummary of the weather conditions during the observation series shown in Figures 12.1 and 12.4 (Swedish tests) Observa- Relata- tion Nature Atm, Wind tive series | of current Temp. pressure m/s humidity Precipitation, etc. No. °C % 1 - + -12 993 8 81 Heavy snow storm 1 -12 994 8 81 a ” ” 1 + -13 994 8 81 » oo» ” 2 + -1l 990 3 83 Abundant fine-grain snow 2 -9 990 3 83 ” os ” 3 + -— 8 1021 9 85 Light snow fall 3 -— 8 1021 9 85 » oo” ” 3 - -7 1023 6 87 Overcast, no precipitation 4 ++ +2 1017 7 80 Clear sky 4 +3 1018 7 77 Partially overcast 4 - +1 1019 1 72 Clear 5 + +4 988 9 Overcast, no precipitation 5 +4 987 6. ” ” ” 5 - +3 983 10 90 Rain 6 - -5 984 9 91 Fine dense snow 6 -5 984 10 90 ” oy ” 6 + -— 4 985 10 90 » » ” 7 + —-14 998 2 80 Clear, hoar frost on the cable 7 -8 999 0 71 Clear 7 - -9 999 1 71 ” Ea References (6) (8) (9) (30) (58) ($8) to (99) inclusive (101) (103) (111) (125) to (169) inclusive 13.1 Problems of high-voltage, heavy-current valves It is well known that high-vacuum, hot-cathode valves are capable of withstanding high voltages but also have the inherent property of a large space-charge drop. The space-charge is due to the accumulation of electrons between anode and cathode and tends to prevent further einission of electrons from the cathode. . To reduce the space-charge drop, a small amount of inert gas, such as argon, is introduced into the valve, its ionisation providing positive ions in sufficient quantities to neutralise the -large electron space charge and thus reduce the voltage drop of the valve. Although this practice increases the efficiency, it is detrimental to the inverse voltage withstand capacity of the valve. Hot-cathode valves, with or without gas, require a large power consumption for cathode heating, and considerable cathode heating time must be allowed to elapse before switching these valves into service; a final difficulty is that the persistent removal of material from the cathode because of ion bombardment severely reduces the life of solid cathode valves, y With mercury are valves the ionisation of mercury vapour neutralises the space charge, as ‘in the case of a valve using inert gas, and thus a very low voltage drop is obtained. The (cathode spot, formed on the mercury surface, has an almost unlimited capacity for efficient ‘electron emission and hence very large current outputs can be achieved. With normal ‘ambient temperature, non-ionised mercury vapour has a large voltage breakdown thres- (hold. Rectifiers based on semi-conducting materials, and capable of handling large powers have also been developed. They have a high internal voltage drop at high voltages, however, and the need for grid control is decisive in making the mercury arc valve the only suitable type, at the present time, for converting a large amount of power for the purposes of h.v.d.c. transmission. But it would be unwise to preclude the possibility that controlled semi- conducting devices, such as the controlled silicon rectifier, may have important application in this field in the future. The superiority of the bridge-circuit arrangement of valves has been discussed in Chapter 2. This form of connection requires the use of single-anode valves, or multi-anode valves with anodes connected in parallel to a common phase, so that they work in a similar fashion to the single-anode valves. CHAPTER 13 High Voltage Heavy Current Mercury Arc Valves penn rene ul Wits steed 1a - 2 - 2 hoe Shia en i 213 Figure 13.1 Schematic diagram and photo- graph of a Russian valve rated at 130 kV, 900 A (max.) designed for Stalingrad- Donbass transmission system a i Take Tn 7 EY 214 (1) anode: (2) cathe 3) grids: (4) MERCURY ARC VALVES SSS iy SS a Se Figure 13.3. A German high-voltage, single-anode Prototype valve for 100 kV, SOA. Cap insulator Principle with long internal and external creep-paths. A Anode heater within graphite body. Intermediate potential electrodes (graphite). grid. D Main de-ionization baile Figure 13.2 Schematic dia- gram of a Russian valve rated at 120 to 130 kV, 150 A (max.) used in the experimental Kashira-Moscow transmission anodes of the ign : (5) upper system exciting anode: (6) 2 (1). shields: (8) hood; (9) cathode le: (10) anode . insulator * By, Bs, By C Control ee — 5 } a pry — — Tey o~- oe oo om es ul ease el satelite tas MERCURY ARC VALVES 215 : 3 & i a, a i itt cise Figure 13.4 A Swedish valve for the English Channel project rene ne RN RN rm "rT 1 won 3 [ [ [ a 216 MERCURY ARC VALVES In normal industrial practice, mercury arc valves have been used for high currents, low voltages, i.e., of the order of current in kilo-amperes at voltages below 1 kV for elcctro- chemical industries, and at high voltages, low currents, i.e., of the order of tens of kilovolts, and tens of amperes, for radio transmitting equipment. Medium-voltage high-current valves, i.e., 1.5 kV to 3 kV, and about 1,000 A, have also been developed for railway traction. None of these ranges of valves are suitable for h.v.d.c. transmission which requires high- voltage, high-current valves of the order of, say, 100 kV and multiples of 100 A. The mercury are valve has always presented difficulties, since there is an insufficiency of ‘knowledge about the mechanisms of the arc and the cathode spot; this has resulted in far reaching empirical development svithout_a_great_deal_of theoreticaLunderstanding of the processes involved. Thus although a practical approach exists for the design of high-voltage, high-current valves, much fundamental work remains to be done. Success in manufacturing such valves has been achieved by the use of a graded anode struc- ture, which consists of several electrodes between which the total voltage is divided and evenly distributed. The cathode spot, furthermore, has so many varied characteristics, and measurements on it are so difficult that no theory has yet been advanced which can explain it completely and convincingly. Figure 13.1 shows a Russian valve rated at 130 kV, 900 A, designed for the Stalingrad- Donbass transmission system."25) Figure 13.2 shows the schematic diagram of a Russian valve rated at 130 kV, 150 A used in the experimental Kashira-Moscow transmission system. “26) Figure 13.3 shows a German valve ‘S8) rated at 100 kV, 50 A, for experi- mental h.v.d.c. use, and Figure 13.4 shows a Swedish valve, of rating 105 kV, 800 A maxi- mum, for use on the Channel project.53) It can be seen that the valves above vary a great deal in their detailed design, mainly on account of the empirical way in which they have been developed. In the following chapter an attempt will be made to explain the principles and requirements of the various compo- nents of valves for h.v.d.c. converters. 13.2 The mercury yapour are 13.2.1 Operational features of a mercury are discharge A mercury vapour are is a cold cathode are in which conduction mainly takes place because of the “emission” of electrons from the cathode spot, “ionisation by collisions” of mercury vapour and, finally, collection of electrons by the anode.’ It partly takes place by virtue of the collection of positive ions by the cathode. When an electron, which has been accelerated by about | volt, strikes an atom it passes - through the atomic structure and tends to be deflected with negligible loss of energy. The next stage is reached when the electron energy is raised to the level required for the atom to be in its lowest excited state and the clectron’s energy may be transferred to the atomon collision. : The excited atom, after its lifetime of about 10-8 seconds, loses its energy by the emission of light, the wavelength of which corresponds to the energy level of the excited state. The potential required to accelerate an electron to a sufficient velocity to excite an atom in this way is called the “excitation potential”. ’ ‘ m. ere cer Eo THE MERCURY VAPOUR ARC 217 Atoms also exhibit what are called “long-lived metastable states” from which they do not fall so readily by the radiation of light. Such atoms tend to remain in this particular type of excited state until they are de-excited by wall collisions or are transferred to an ordinary excited state through encounters with electrons or other atoms.927) There may be a number of such states of excitation. The highest excitation state is the “ionisation state” which may result if the impinging electron has sufficient energy to cause the ejection of an electron from the atom, thereby yielding a positive ion and two slow electrons. The corresponding electron potential is called the “ionisation potential” and is 10.39 V for mercury vapour. Similarly, there will be further stages of ionisation in which an atom is deprived of two, three, four, etc., elec- trons. Transfer to any of the above states can take place directly in one encounter if the electron energy is sufficiently high. When gas molecules, as opposed to gas atoms, are bombarded by electrons the results are similar in principle but differ considerably in detail on account of the extra possibilities of molecular vibration and dissociation. : Jonisation can also result from the impact of positive ions, fast moving atoms, and excited (particularly metastable) atoms, etc., although the likelihood is small compared with ionisa- tion by electron collisions. Since only a small voltage drop occurs across the discharge in mercury arc valves, this low accelerating voltage makes it uncertain whether the ionisation occurs mainly by impact of electrons with normal atoms, or with atoms already excited by previous electron collisions. The former is called direct and the latter cumulative ionisation. To understand the rectifying action in a mercury arc valve, consider an envelope consisting of two electrodes at a distance / apart and filled with some gas (Figure 13.5) at low pressure say, below 1 mm of mercury. If a small voltage is applied, a current with a density of the order of 10-8 A/cm? will flow due to the small amount of ionisation which is produced continuously by cosmic rays and the earth’s radioactivity (2 to 5 ionisations/cm/sec) which also causes a small amount of electron emission from the negative electrode. This discharge is the dark “Townsend discharge”; it increases with increase in voltage and the current flowing is known as the “precurser discharge’. As the applied voltage is increased, a level is reached when the discharge suddenly changes into a bright glow called the “Glow Discharge”. The voltage across the valve now falls considerably witha rise in current. The cathode to anode region"!25) consists of (Figure 13.5): (a) The Aston dark space, of very small length. (b) The cathode glow, a thin glowing layer adjacent to the Aston dark space. (c) The Crookes, or Hittorf, dark space. (d) The negative glow, a glowing region which progressively darkens in a direction to- wards the anode, and finally becomes completely dark. (ec) The Faraday dark space, several times longer than Crookes’s dark space. (Y) The positive column, or plasma, which may also consist of alternate bands of bright and dark spaces. (g) The anode glow. (4): Finally, the anode dark space of very short length immediately adjacent to the anode. The distance from the cathode to the end of the negative glow (of length 8) is termed the on 1 oT = eels . ail x oxic 7 ‘east a 218 MERCURY ARC VALVES Crookes or HittorF dark space Cathode glow / Negative glow Aston dark Faraday dark space vA space é 2 ! Positive column wf or plasma Anode glow ? Anode dark| space Figure 13.5 Glow discharge phenomenon “cathode fail space’’, and it is across this space that most of the voltage drop (which may be a few hundred volts) appears. Compared to this drop, the drop across the positive column and anode is negligible. A simple, yet adequate interpretation of the glow discharge phenomenon is as follows: Electron release from the cathode takes place because of positive ion bombardment of the cathode. ‘ The electrons which are just leaving the cathode are slow at first and as they gather speed, a layer of excited atoms appear (the cathode glow above the Aston dark space). As the electron speed increases, the excitation probability passes through a maximum and then falls whilst the ionisation probability is still increasing toa maximum. Thus another dark space (Crookes) appears. The avalanche-like increase in the free electron density implies that soon these electrons will be able to carry the bulk of the discharge current by moving at lower speeds towards the anode. Thus there develops a region of intense excitation and strong ionisation (negative glow) followed by a dark region (Faraday dark space) where little excitation or ionisation occur as the electrons, although dense, have only low velocity. The Faraday dark space marks approximately the terminal point of the ordered beam of electrons from the cathode and the negative glow is the most substantial source of the Positive ions which directly strike the cathode and maintain the necessary release of elec- trons; these electrons, in turn, are essential for producing the ionisation. Since the ions which have been created in the Faraday space have very low speeds, there exists an excess of ionic charge and ina similar manner close to the cathode (Aston dark space), there may be an excess of electronic charge. = [ i | i { \ [ [ [ [ t | [ [ L L a THE MERCURY VAPOUR ARC 219 The positive column or plasma consists of equal densities of positi together with a large number of atoms and molecules in both excited an aancied states. Near to the anode the electrons are accelerated and hence create some excitation, which then reduces in the region very close to the anode where the anode sets up an electron space charge. For small glow-discharge currents (limited by external resistance) the glow over the cathode is confined to a small area in which the current density decreases with decrease in pressure. As the current rises, which will happen with only a small increase in voltage across the discharge, the cathode glow area increases until it covers the whole cathode area, after which, the voltage across the discharge starts to rise rapidly (most of it across the cathode-fall space); the discharge in this region is called “abnormal glow discharge”. As the voltage increases further, the stress on the cathode will increase to such a point that the cathode starts to emit a large number of electrons; complete breakdown occurs with a sudden rise in current accompanied by a fall in the voltage drop across the gap, and the arc is struck. The arc discharge consists also of cathode-fall space, plasma and anode dark space, although in this case the cathode-fall space is very much shorter in length. The arc becomes concentrated to a small bright spot of very high current density and is able to emit a large number of electrons. It is clear that if the current is not limited sufficiently by the external resistance at the first glow-discharge breakdown, an arc discharge will follow. In the glow-discharge the length 6 of the cathode-fall space remains constant under conditions of constant pressure, irrespective of the total gap length being decreased or increased. With increase in pressure, the cathode-fall space decreases, and vice-versa, and approximately follows Paschen’s law: ‘p . §=constant [Figure (13.5)] The value of the constant in this relationship is about 0.7 for graphite electrodes, with & in cm and the mercury vapour pressure, p in mm. The glow-discharge continues to take place as the total gap length is reduced without any change in the length of the cathode-fall region, but with decreasing length of plasma and a corresponding decrease in the voltage drop. : However, when the gap length is reduced, eblow the critical length 8, the glow discharge vanishes. This occurs because of insufficient mercury atoms being-available for collision with accelerated electrons, in order to maintain the necessary ionisation for extracting elec- trons from the cathode. If the voltage is now increased, complete breakdown (arc) will finally occur but without going through the glow discharge phenomenon, although the dark ‘‘precurser” discharge (here called the “inhibited discharge’), does take place before the breakdown. . The breakdown appears in the form of bright spots on the emitting electrode. The impor- tant thing about this critical length 8 is that within this range, the smaller the length of the gap the larger will be the breakdown voltage. Any decrease in pressure will increase the critical length § and consequently increase the breakdown voltage of any gap of given dimensions. ; It is in this region, where decrease in pressure and decrease in distance between the elcc- coy em cea am pom sey i 220 MERCURY ARC VALVES trodes increases the breakdown voltage, that mercury arc valves work. This principal also applies to the gaps between the intermediate electrodes. It is clear that the voltage of a valve without intermediate electrodes cannot be increased by any great amount because of the necessity of keeping the anode at a considerable distance from the cathode for protection from vapour jets, etc. (this is discussed further in Section 13.5). It is necessary that the valve should fire at very low, positive anode-cathode voltages and since a very high voltage, of the order of several kilovolts, is required to strike an are, it is necessary to initiate the arc by some means. It so happens that the cathode spot continues to emit electrons only as long as the vapour above it is sufficiently ionised, and hence as long as sufficient current is sustained by the vol- tage acting through the external impedance. This limit is about 3A for a free spot. Between 3 A and 6 A, the spot is likely to be extinguished at random, although the rate at which it extinguishes decreases with increase in current and disappears altogether for currents above 7 A. Thus to maintain the are, the total cathode spot current (which may be due to several anodes in the valve) should not be less that, say, 6A. Above this current, the arc is self maintaincd and a very large number of electrons (practically unlimited) can be obtained from the mercury cathode. The necessity for igniting the arc and also maintaining the minimum stable cathode spot current can be met by the provision of additional electrodes. The ignitor is an electrode which is brought into contact with mercury, the contact then being broken to initiate an arc, which is then picked up by the excitation electrode; this latter electrode is arranged to carry sufficient current to maintain a stable cathode spot on the mercury surface (dis- cussed further in Section 13.10). If the cathode spot is initiated by such means as these, and some of the ionisation so produced can diffuse to the anode, then the valve can fire and pick up current at low positive anode voltages. Control arrangements can be introduced by the provision of a negative grid electrode which prevents any ionisation from diffusing into the anode region, and thus prevents the positive anode from picking up current, until the grid is made positive (Section 13.4). Thus when a valve is conducting, the discharge consists of the cathode spot (with a cathode-fall space) which emits the required number of electrons and carriers the total anode and excitation electrode current, the plasma extending from the cathode-fall space to very near the anode and consisting of approximately equal numbers of electrons and positive-ions along together with a large number of neutral atoms and molecules of mercury vapour (some in excited states), and finally, very close to the anode, the anode sheath where electrons predominate over ions. When the mercury valve is not conducting, the discharge consists of the cathode spot emitting clectrons corresponding to the excitation current, plasma extending to a region near the negative grid which is covered by a sheath of positive ions and serves to prevent the infiltration of electrons towards the neighbourhood of the anode. 13.2.2 Voltage drop across the are The total voltage drop across the are, which is of the order of tens of volts, is made up as follows: : [ THE MERCURY VAPOUR ARC 221 iLalso i (a) The anode voltage drop ofa The anode voltage drop is a function of current density, ‘atcury Vazour pressure and 0 of Ely anode geometry and material, because these factors affect ts positive ion and electron 2ction ‘ densities immediately before the anode. The drop is usually »ss than 1% volts. In some special cases it is negative, e.g., with a hollow, tubular anode. showing :zat sometimes the positive ion density swamps the electron density near the ano<=, (6) The voltage drop in the plasma The plasma voltage drop, varying between 0.02 to 0.2 V, ca, represents the necessary expenditure of energy to compensate for the deionisation whi: is taki place and hence is needed to maintain the necessary ionisation for a particular luce of current in the valve. (Deionisation phenomena is discussed in Section 13.3.) | (c) The cathode voltage drop As mentioned before, the glowing plasma of the arc dischzrge is separated from the cathode by a short region called the ‘cathode fall space” or “cz-hode zonz” through which the electrons emitted from the cathode spot are accelerated ‘+ sufficient speeds to cause ionisation of the mercury vapour. In the case of liquid mercury cathodes, the mechanism of cle<tron emission and the con- stitution of the cathode-fall space are not fully understood. J: is across the cathode-fall 7 space that a significant part of the voltage drop occurs, and t:+ plasma boundary above forms what is frequently regarded as a virtual anode. Until recently, the figure for cathode voltage drop was take+ to be approximately29) 10 V, but recent measurements have shown that 8 V is more ;ealistic?2 under normal \ conditions. Other measurements also seem to indicate that the cathode voltage drop varies with pressure, being about 11 V at 1 micron, and 8 V at zbout 100 microns.(9) A contrary suggestion has been made recently that this volt: drop in the region of the cathode space charge may be 20 or 30 V or even more, whilstti. total potential difference across the arc may be less than this; this view, if correct, would imply a negative voltage drop, and would arise from the positive ion-sheath situated very close to the cathode and an electron cloud above it."93)) 13.2.3. Characteristics and mechanism of the cathode. spot The most striking feature of the mercury arc valve is the manner in which the electrons are obtained from the cathode with such a small energy consumption, the source being the bright cathode spot, which moves erratically on the mercury surface, giving out intense jets of vapour. Many theorics, methods of calculation, and measurements have been put forward to explain the maintenance of this phenomenon and its various characteristics. Great dif- ferences exist between the measurements of various authorities mainly due to the erratic motion of the spot and the agitated mercury surface, and variants in the characteristics have been obtained in the results from different experimental devices. This subject is extraodinarily complicated, and a very extensive literature exists; various theorics and characteristics of the phenomena can only briefly be discussed here. PS NEI Taal eal L [ Ca C2 3 ark co (4 CoD wan 222 MERCURY ARC VALVES 13.2.3.1 CURRENT DENSITY AND SIZE OF THE CATHODE SPOT The whole of the available current passes through one or more spots, although it is doubtful whether the whole’ area of a spot contributes to its associated current and whether the current distribution is uniform. The generally accepted value for current density in the spot is 4,000 A/cm?2,5) although suggested values of 2.105 A/cm? and 108 to 107 A/em2 concentrated over small areas are also accepted in some quarters,(96) Whatever may be the density, the spot size increases with the current. The current of a single spot is known to be between 3 and 15 A, with great probability of 6 to 7 A being the real figure, and is concentrated on the bright spot area witha perpendicular distance of about 0.002 cm above and 0.00002 cm below the surface of the mercury, and with a diameter lying between 0.034 and 0.07 cm. (91) (98) Von Bertele has suggested®®) that the spot density actually varices, the different values above being obtained at different times after the intiation of the spot; this view would imply that the spot has a life cycle, starting with a current density of the order of 107 A/em2, after which it expands, the current of the spot then being more or less constant at an average value of 5A. The density reduces with time until the spot divides into smaller parts or, more often, shortly before breaking up a new highly concentrated spot is created which recommences the cycle anew. But it would still be correct to consider 4,000 A/cm? as the average spot density, as the life time at high density is considerably smaller than at low density. Thus the number of spots and their sizes may vary continuously and may partly account for the continuous movement of the spot. 13.2.3.2 SPOT TEMPERATURE : The spot temperature is high due to the cathode losses occurring at the spot. The pool also acquires an average temperature (about 50° C for power valves) depending upon its size, current and cooling arrangement. On this average temperature is superimposed the temperature peak at the spot, although the spot itself may have different temperatures in different regions. Generally accepted values of spot temperature vary from 200 to 400° C, (92) (98) although temperatures below 200° C3) and up to 2,000° C have also been suggested. 13.2.3.3 VAPOUR JETS AND THE TANBERG EFFECT This is a characteristic of a free cathode spot by which jets of mercury vapour are ejected by the cathode spot. This was first noticed by Tanberg't32) who gave vapour velocities of the order of 15x 105 cm/sce. It is generally believed that the jets are formed by the pressure generated by the continuous evaporation of the electrode material inside the spot. It has also been suggested that the jets are formed by the “kicking out” of electrode atoms by the ions attracted by the cathode, in a similar mechanism to cathode sputtering. Any theory to be fully satisfactory has to explain the stoppage of the vapour jets when the spot is anchored (Section 13.17). Various measurements show an average jet specd of 10% cm/sec, and’ the reaction of the jets on the pool surface excites irregular surface waves which may attain amplitudes up to 12 mm.©2 The jets change direction as the cathode moves across the pool surface. r [ [ et is aether si in sm? iwofa ifthe out meter fo: would THE MERCURY VAPOUR ARC. 223 13.2.3.4--ERRATIC MOTION OF THE CATHODE SPOT The spot moves erratically, dancing over the pool surface. Duc to the tremendous pressure on the spot, the mercury surface in the region of the spot is depressed; the spot itself docs not stay in the depression, but moves up the sides whereupon the pressure is applied to new points, the spot moving swiftly across the mercury surface with many erratic changes in direction. The speed of movement is considered to be about 100 cm/sec, although varying speeds ranging from 1 to 1,000 cm/sec at random have also been suggested. Erratic motion has also been attributed to the force exerted by the vapour jets emerging from the spot. 13.2.3.5 RETROGRADE MOTION'!30) This is an electromagnetic property of the cathode spot in a low pressure mercury arc in that the cathode spot will move in a direction contrary to that for normal conductors when placed in a transverse magnetic ficld. Measurements have. shown that the retrograde velocity rises linearly and rapidly up to a field of 3.10° A-T/m, and then levels off at a speed of about 110 m/sec. Then at about 11.105 to 15.105 A-T/m the velocity increases suddenly to about 1.8 times, then increases more gradually and then becomes more constant. Finally, at a field of about 21.103 A-T/m, there is again a similar rapid increase in the speed. An increase in current increases the retrograde velocity, an increase in pressure reduces the retrograde velocity so that eventually, the spot comes to rest and will then move in the forward direction. Similarly, the introduction of inert gas reduces the speed and eventually the spot moves forward. There are several theories put forward to account for this phe- nomena but they are beyond the scope of this book. 13.2.4 Theories of spot emission (a) Thermionic emission This is one of the early theories, in accordance with which the local temperature inside the spot region must be high enough to cause thermionic emission; it was suggested that the positive ions, in being attracted towards the cathode, bombarded it and caused heating to a temperature suflicient for thermionic emission. Furthermore, it has been shown that thermionic emission theory fails for those cathodes with a boiling point lower than that necessary for thermionic emission.©® At a pressure of 760 mm of mercury the boiling point of mercury is 357° C, far below the temperature required for thermionic emission. (6) Field emission This theory was first suggested by Langmuir. The positive ions, which are drifting by attraction toward the cathode, give use to such a thin layer of cathode space charge that an cnormous potential gradient occurs along its length. This gradicnt was considered to be so large that it could draw electrons out by the so-called “cold cathode effect” or “auto- electronic emission”. About 105 V/cm is required for the process of field emission, and as it was believed that the cathode drop was about 10 V,.the thickness of the space charge film for satisfactory emission by this process would have had to be about 10-6 cm. A number of measurements have shown, however, that the thickness is considerably more than 10-6 cm, and since the voltage drop is about 8 V the field intensity at the cathode is clearly less than that to produce field emission.233) On the other hand, on account of nr i Te RN RR — > > ae 224 MERCURY ARC VALVES molecular roughness of the surface, the true intensity at a few points may well be much more than these elementary calculations indicate. It has also been suggested that the field is caused by the ion charge accumulated by an impurity layer on the cathode surface. Destruction of the impurity film at random would also account for erratic movement of the spot.431) Apart from the high field intensity and high current density required for field emission, a voltage of 4.53 V (the work function of mercury) is required for emission of electrons, in addition to the ionisation potential of 10.3 V which is required if the direct ionisation of mercury vapour above is assumed. Recent measurements have shown the existence of a ficld of 6x 106 V/cm, and it has been suggested that the voltage drop across the dark space must be at least 20 V.(130) (c) Photo-electric emission The production of electrons by photo-electric emission has been criticised on the grounds that it is not efficient enough to account for the high current densities encountered. But the spot is known to be an intense source of ultra-violet radiation, and being so close to the cathode surface, the contribution of photo-electric emission in the emissive phenomena of the spot might well be greater than expected. (d) Theory of excited atoms : According to this theory, first suggested by Compton"3) and recently developed by Robson and Von Engle, 35) the electrons are emitted by the impact of excited or metastable atoms which arrive at the mercury surface. Impact of metastable atoms can cause the ejection of electrons from metals and every incident, metastable atom can eject 0.1 to 1 electron provided that the energy of the atom exceeds the work function of the metal. Mercury has two low energy and easily excited states (4.66 eV and 5.46 eV) which are metastable, both having greater energies than the work function of clean mercury (4.53 eV). The temperature of the spot gives rise to a high evaporation rate from the mercury surface inside the spot, resulting in a layer of very high pressure (about 10 atmospheres) mercury vapour, and most of the evaporating atoms are scattered back to the surface in an excited state by this vapour of very high density. Excitation is so rapid that most of the returning atoms (possibly 109 per cent) are excited; these excited atoms are driven back to the cathode and cause clectron emission. This theory has been criticised on the grounds of its unproved assumption of very high vapour pressure at the cathode spot and high rate of evaporation. It has been shown that with appropriate temperature control and cathode cooling the mercury evaporation can be reduced to almost zero (Section 13.17). (e) Combination of theories Besides the theories above there are others, all of which have been criticised mainly on account of the low energy associated with electron emission from the cathode spot. How- ever, it is possible that the various suggested phenomena may be operative together in some combination. Von Bertele®) in suggesting a life cycle of the cathode spot has put forward the view that, in the initial stages of the spot when the current density is high, field emission plays the dominant role but as the spot expands and the current density becomes too low for field emission, another mechanism takes over. His experiments on cathode spot anchoring and on film cathodes have cast a new light on the emission mechanism, and this is dealt with in Section 13.17. om pte at [ DELONISATION 225 more With similar views in mind, Feinberg®35) and his associates"137) are carrying out investi- [ gations into the role played in emission phenomena by the two processes of field emission hBan and photo-electric emission acting in concert; this is based on the fact that application of vould a strong electric field reduces the effective work function of the mercury composing the rad cathode, and so enhances the photo-electric emission. [: It has yet to be shown conclusively how, and under what conditions, any of the above ial of theorics fit the propertics of the cathede spot; there is the probability also, that all of its : properties are not yet known. fe It is important that further work should be carried out and a thorough understanding of the are and cathode spot established; this could contribute to the development of simpler and better valves, in which the difficulties of valve design are not necessarily overcome by f° empirical experiment, expensive auxiliaries, and bulky valve sizes. PELE 13.3. Deionisation(39) o the recombination of electrons and positive ions, is taking place. When a valve ceases to con- duct, its anode becomes negative and then attracts positive ions which strike the anode, lose their velocity and kinetic energy, and tend to become neutralised by recombination at the anode. This constitutes the sniall negative current, similar to the positive current except that the emission of electrons from the negative electrode is now negligible and, in conse- C During both the conducting and non-conducting periods, the process of deionisation, or 2 La 2 the quence, ionisation is decreased by the negative conduction. ‘1 Free electrons in an ionised gas possess translational energies; when one of these electrons comes within the range of attraction of a positive ion it would be expected to move past the : ul ion ina hyperbolic orbit. But permanent attachment involves settling down into an ellip- Ce . tical orbit which can occur only if the electron loses a part of its energy. The only way of losing energy is by collision with a third body during the interval of close approach; clearly this is improbable at low gas pressures. Combination can also take place in the gas by collision of electron and positive ion, the {i: energy dissipation taking place in the form of radiation, this serving the purpose of a third body. But the observed weakness of the radiation outside the regular line spectrum makes ; this process of recombination also insignificant and it has been shown that the light emitted y by the discharge comes principally from excited atoms. Recombination in gas also takes place by two successive collisions instead of collision of three bodies at a time. In this case the electron striking a neutral atom will attach itself a i to the atom and form a negative ion; thercafter, if the negative ion collides with a positive a ion the electron is transferred, any excess energy being carried away by the neutral molecule. ee. Duc to the very low pressure in the mercury vapour valve, this kind of recombination will ome also be very infrequent. 7 He Recombination on surfaces is the most important, in which the surfaces within the valve n act as a third body to absorb the excess energy and momentum. Electrons and ions need low not arrive simultaneously at a point on the surface; one of them may arrive first and remain ‘got on the surface as a surface charge until the other arrives to combine with it. Dust particles, [is sub-microscopic particles sputtered from electrodes, insulator surfaces, floating electrode surfaces, baflles, tank walls, etc., all act this way. iT OFT? fm cs FO as 226 MERCURY ARC VALVES Duc to the above processes, deionisation is also taking place when the valve is conducting and this loss of ionisation from the plasma is compensated and balanced by a corresponding ionisation process which contributes to the small voltage drop across the plasma. The importance of deionisation with regard to the voltage and current capacity of the valve is discussed in later sections. 13.4 The grid The control exercised by the grid over the discharge between the main electrodes may be explained by Langmuir’s theory‘89 (29) of a sheath surrounding an electrode inserted in a neutral plasma. ! Ifa probe (or electrode) is inserted in the neutral plasma of a discharge between anode and cathode, it will be surrounded by a sheath of electrons or positive ions depending upon whether the probe is at a positive or a negative potential with respect to the plasma potential at that point, so that at the periphery of the sheath the potential is the same as that of the plasma. Thus the field of the grid is cancelled by the sheath and its influence extends only up to the edge of the sheath. The positive ion current carried by a negative electrode will correspond to the positive ions that reach the outer periphery of the positive ion sheath, on account of their random motions in the plasma. So also will the electronic current carricd by a positive electrode. But due to the small mass, and consequently high velocities, of electrons compared to that of ions, the electron current density in a uniformly-ionised gaseous plasma is hundreds of times greater than the positive ion current density. If the electrode is at a slightly negative potential with respect to the plasma potential, the positive ion sheath will be small and electrons which have high velocities may still reach the electrode against the retarding field which they encounter within the positive ion sheath and the current carried will be the difference of electronic and ion flow. If the potential is made more negative, less and less electrons will be able to penetrate through in this way, and a point will be reached when electronic current is equal to ion current (i.c., the total current is zero) and the electrode potential is then referred to as the “Langmuir potential”; beyond this potential, the positive ion current will be in excess of the electronic current and ultimately the current would be due entirely to positive ions. Cn account of high electron velocities an insulated surface or a floating electrode will acquire an clectron charge sheath and hence a negative potential; equilibrium is reached when large numbers of electrons are repelled by the electronic charge and those reaching its surface are equal to the positive ions also reaching it, the net current being zero. ; The greater the potential difference between the electrode and the plasma, the greater will be the sheath thickness. Also, the greater the ionisation density of the plasma, the smaller will be the sheath thickness, for under these circumstances a smaller sheath thickness will be able to neutralise the potential difference. ! When the valve is non-conducting the ionisation between grid and cathode is that corres- ponding to the excitation current, and the positive ion sheath in front of the negative grid is such that it completely overlaps the holes or passages through the grid electrode. The negative grid carries some positive ion current depending upon the positive ions which PRINCIPLE OF HIGH-VOLTAGE VALVES 227 diffuse towards the grid and reach the sheath surface, and thus depends upon the grid voltage, exciter current and distance of the grid from the exciter. If the grid is now made less negative a point will be reached when some electrons will be able to penetrate through the grid into the anode region where they will be accelerated (if the anode is sufficiently positive) and produce more electrons and ions. The ions in turn help to counteract the negative field of the grid, the sheath thickness is reduced and more electrons are able to penctrate through the grid. This effect is cumulative. The ionisation f increases with increasing anode current and grid completely loses its control. If the grid is made more negative again after the firing has taken place it cannot regain control. This is because the ionisation density which is present is very great and consequently the sheath [ thickness is very small. In order to achieve successful firing of the valve at very small anode voltages the grid must be made sufficiently positive with respect to the cathode. The positive grid attracts rial electrons from the ionised plasma and carries electron current, thereby increasing the ionisa- tion in front of it. The current is limited by external grid resistance. The electrons then diffuse through the grid channels and produce ionisation on the other side of the grid. The electrons are further accelerated towards the first intermediate electrode, which carries some [. electron current and increases the ionisation in front of it. The electrons further diffuse through its channels into the next gap and finally ionisation reaches the anode, which draws a greater and greater number of electrons to increase the ionisation very quickly indeed. Thus in order that the ionisation can successfully diffuse through the extended grid and i electrode channels and the valve can successfully fire at a low positive voltage, it is necessary that the grid and electrodes (through the voltage divider circuits) should carry sufficient he current when they are made positive and are thercby enabled to produce sufficient ionisation ji: in their respective neighbourhoods. Apart from a dependence on constructional features of the valve, the grid voltage at which a valve fires depends upon the anode voltage, vapour pressure, ionisation, etc.; [« because of the varying nature of these factors, it is desirable to apply a steep positive pulse in order to ensure correct firing of the valve. Also due to the vapour blasts which carry intense ionisation, the local vapour pressure and ionisation density is liable to great varia- of tion; in consequence, the grid is made dimensionally large in the direction of the discharge | so as not to lose control. will For invertor operation, the grid needs special consideration. When a rectifier valve is | blocked, the anode is made negative with respect to the cathode for the greater part of the [is blocking period; hence its electric field assists the grid. In invertor operation, when the valve stops conducting, the voltage first becomes negative and then after a few degrees of the will cycle changes to positive, and although the grid may have successfully separated off the er plasma, the high, positive anode voltage tries to contract the grid sheath. Thus to enable Ee the grid to block successfully for high values of positive anode voltage, it has to be made of considerable length. In most high-voltage valves two adjacent grids have been used. fi 13.5 Principle of high-voltage valves with voltage division The In accordance with section 13.2.1, a discharge and consequently an arc occurs when any bich positive ions which may be present, are attracted towards the negative electrode and have Orel 228 MERCURY ARC VALVES Seed: Gee) suflicicntly high velocitics on reaching the electrode surface to liberate a necessary mini- mum number of clectrons; these electrons, in turn, are accelerated toward the positive clec- trode and undergo sufficient collisions to produce more positive ions than originally existed in the inter-electrode space, ete. . When a negative voltage is applied between anode and cathode in a valve with no voltage division, most of the voltage drop appears across a thin layer (the Langmuir sheath) near the anode; any positive ions which arrive near the anode are accelerated through this thin sheath and acquire sufficient speeds to cause electron emission from the anode. If, by means of a voltage divider, an even voltage distribution is obtained throughout the length of the inter-electrode space, then positively charged particles will be accelerated through the same voltage but will encounter large numbers of other particles; thus the gain in energy from the accelerating voltage will be offset by the loss of energy through collisions, and the speeds attained will be insufficient to cause electron emission from the anode. Also, through the great number of collisions with other particles, there is increased likeli- hood of deionisation when ions reach certain of the surfaces within the valve. Thus the effect of spreading the voltage is as if the charged particles were accelerated through only a fraction of the total voltage. This principle was invoked“40 jn an experi- mental valve built by ASEA (Sweden) in 1933, and is shown in Figure 13.6; this valve has a long tube of semi-conducting material inside a glass anode tube, with the intention of providing a continuous distribution of voltage over the whole length of the semi-conductor tube, and permitted a substantial increase in voltage without are-back. Two modifications with different lengths of semi-conductor inserts indicated that the admissible voltage increased with the length. Semi-conductor tubes had a short life and were replaced by intermediate electrodes and external voltage dividers in order to distribute the voltage along the whole length of the anode structure. Any charged particles which are accelerated in a gap lose their energy by collisions with other particles and the electrode surfaces, where they may even be neutralised. Thus if charged particles lose all the energy gained in transit through the voltage of a gap, then the total voltage that a valve can withstand will tend towards the safe, voltage-withstanding capacity of each gap multiplied by the number of gaps. The voltage-withstanding capacity of a gap can be increased by decreasing the pressure and the distance between the electrodes. Apart from these two factors, ionisation plays a very important part. When a valve ceases to conduct, there will be considerable ionisa- tion left in the gaps which will greatly decrease the breakdown voltage of a gap and also greatly amplify the pre-breakdown “precurser” discharge; this latter may even increase the ionisation to the extent of causing a breakdown. Figure 13.7 suggests how the breakdown voltage of a 2 cm gap may be expected to depend on different vapour densities and ionisa- tion.69 Tt can be seen that an increase of ionisation greatly reduces the breakdown voltage of the gap; mainly because of impurities (see Section 13.1 5), the breakdown voltage is likely to be much less than given in Figure 13.7. The gap length cannot be reduced much below 2m, on account of the possibility of short-circuit caused by accumulated sputtered material. [ 13.6 Arrangement of electrodes Three basic requirements undcrly the construction of an intermediate electrode: Serene erence renee (oa - 7 eb apeartentalis Sle etecte Sat ON a ai [ ARRANGEMENT OF ELECTRODES 229 nini- (i) It should impose its potential by breaking the continuity of plasma through its {: channels as soon as a negative voltage is applied, i.e., its channels should be long and narrow. (ii) It should have sufficiently wide channels to allow the necessary ionisation to reach the anode by diffusion, in order to obtain successful firing for a small anode voltage as soon as the grid is made positive. (iii) The free passage through an electrode should be sufficient for the current rating of valve in order that the current density should be within reasonable limits. it In spite of local interruption of the continuity of the plasma when the valve is non- fed conducting, charged particles within the gaps can still be moved through the channels into sain : the anode space. Figure 13.8 (a) shows® the equipotential lines across the. gap between intermediate electrodes with a concentric channel, which shows the curvature of an elcc- trostatic lens which will tend to accelerate and focus the charged particles into the centre line of the channel. The space within the body of the electrode, within which no field exists, allows the par- ticles to pass through the electrode into the next gap where it will be further accelerated and produce unwanted ionisation. This will reduce the voltage withstand capacity and produce rapid destruction of the anode due to the bombardment of high velocity charged particles. Such a concentric arrangement will, no doubt, have a much higher voltage rating than would Grading electrode Figure 13.6 Swedish valve design tested in 1933 This figure shows, in principle, a valve which was tested in 1933. It had a long tube of semi-conducting material inside a glass anode tube, with the intention of providing a continuous distribution of voltage over the whole length of the semi-conductor tube. This acted, therefore, both as a voltage divider and as a set of grading electrodes. This design showed a definite positive effect, as it enabled high inverse voltages to be carried without arc-back ahi ahd iit nan the ne tae, 230 MERCURY ARC VALVES Vo ‘ \ MeN Sh \2 4 7 \ Y 3 8 LT NTS 2 NEAL YN * 4 NP NIN, 10 NIN 2 5 8 1 4 Sparking potential, volts 0 270 40 60 80 00 90 140 160 {180 Saturation temperature to °C aewentire 1 4 1 1 19-4 (9-43 /0-/2 lo-u /Q-19 Vapour density, @£9- mol/cm 3 Figure 13.7 Sparking potentials of graphite anodes in mercury are rectifiers referred to average anode space ionisation 2 and vapour density 0, and a zero potential distance of about 2cm. Average number of ions/cm? = 10 be the case without voltage division, but better overall results can be obtained by arranging the electrodes so that there is a displacement of the centre lines of apertures (channels) in adjacent electrodes as shown by Figure 13.8 (db); this provides mechanical obstruction to the charged particles so that they lose their energy and never attain velocities sufficient to cause substantial electron emission from the electrodes. This arrangement, however, still allows particle acceleration through two successive gaps, as shown in Figure 13.8 (b), and electrodes with twisted fins as in Figure 13.8 (c), or a simple sequential arrangement of cones as shown in Figure 13.8 (d), have been suggested. Although in present valve designs, indicated in Figures 13.1, 13.2, 13.3 and 13.4, electrode arrangements similar to those shown in Figure 13.8 (b) and 13.8 (d) have been provided, such mechanical obstructions to the motion of the charged particles is not desirable, according to Lamm, on account of the sputtering of electrodes (Section 13.13); this action causes long term detrimental effects and it is agreed that it is better to achieve braking of the particles of collisions with gas molecules in free space rather than by collisions with electrode sur- faces, i.e., provide straight concentric channels through the electrodes.“4 Sufficiently high voltages do not scem to have been obtained so far with such a construction, however. — ae ae oe |. — Riis Pi en hae ie et ARRANGEMENT OF ELECTRODES 231 (6) Figure 13.8 Types of intermediate clectrodes (a) Co-axial tubular electrodes ;_ (6) Displaced tubular electrodes ; (c) Tubula# electrodes containing twisted fins; (d) Electrodes formed from conical annular rings a en eet ee iin - ssa 232 MERCURY ARC VALVES 13.7 Voltage dividers for valves with intermediate electrodes’80) It was mentioned in Chapter 2 that the main difficulty in obtaining voltage division with serics-connected valves is due to the large stray capacitance to earth of the auxiliaries connected to the cathode, and that these stray capacitances are so large that to offset their ill effects the voltage dividers become very expensive and uneconomical for more than about four valves in series. Ina multi-clectrode valve with subdividing electrodes [SE in Figure 13.9 (a)] there are no stray carth capacitances except the capacitance C, of the cathode and its auxiliaries, Instead there will be small internal capacitances C,, of the order of 105 »F (depending upon their spacing, diameter, etc.) and internal resistances R,. These internal resistances depend upon the amount of ionisation present in individual gap and may vary in time, and from gap to gap. Because the variable conductance due to the ionisation is generally larger than the capacitive admittance, an external potentiometer is required whose admittances should be relatively larger than the internal admittances, so as not to permit fluctuations in internal admittance to influence the voltage. division. Since the internal admittances are small, however, the voltage subdivision can be obtained by inexpensive voltage dividers, Z, as in Figure 13.9 (a). The voltage dividers may be a resistive chain, Figure 13.8 (4), a capacitive chain, a reactive chain, or the combination of these elements. A capacitive chain connected to electrodes through the resistances, as in Figure 13.9 (c), has been suggested as the best arrangement, since the capacitors have a good transient response whilst the resistors obviate any tendency to parasitic oscillations. Furthermore, these resistors will limit the current to cach elec- trode and thus greatly reduce the effect of voltage breakdown of individual gaps. Non- linear impedances may also be used to obtain an even voltage distribution. 13.8 Transfer of mercury from the pool. Control of temperature and pressure The flow of current in the valve is accompanied by an are drop and consequent loss of power. This loss includes the excitation of atoms accompanied by subsequent light emission, ionisation and acceleration of ions with subsequent dissipation at wall and cathode surfaces, the production of vapour jets and splashing of mercury, and the acceleration of electrons with subsequent dissipation at the anode as heat. Most of the energy released at the cathode is in the form of vapour jets, splashing and evaporation of mercury from the cathode surface and depends on the temperature difference between the cathode surface and the upper regions of the valve; as a consequence of this there is a large transference of mercury from the cathode to the upper regions. In power valves this transfer of mercury may be between 4 and 6 mgm/coulomb. Now the voltage withstanding capacity of the valve during non-conducting periods depends very greatly on pressure. The lower the pressure the higher is the breakdown voltage, although this pressure will have a lower limit in order to provide sufficient vapour for ionisation to carry the required current during condacting periods. It is evident that the vapour pressure inside the valve must be kept well under control and should not fluctuate much about the value fixed for a particular design; for high voltage valves this value will generally be between 0.5 to 5 microns. STD lal la a ba a ene TRANSFER OF MERCURY FROM THE POOL 233 It is thus important that the mercury transferred from the pool to the upper regions should be condensed and returned back to the pool at the same rate, and that control measures should be taken to eliminate the disturbing effects of the jets and hence maintain steady pressure in the anode region. To do this a large condenser chamber (tank) is provided together with efficient cooling of the tank. These arrangements and the process of cooling are wasteful in terms of valve dimensions; compared to the electrical space, which consists of cathode space, anode structure and the necessary discharge space, the inactive space required for mercury condensation and cooling to obtain steady vapour pressure in the anode region may be as high as six to ten times as great.(S8) (oD If means of reducing mercury transfer could be achieved, there would be a corresponding decrease in the size of condensing chamber required. The evaporation of mercury from the pool surface can be reduced by external cooling of the cathode. To keep the average pool temperature low independently of external cooling, a sufficient surface area and a i HHPAR ~~~ 5° ~~~ HHH Figure 13.9 Voltage dividers for a valve with intermediate electrodes a OS arene ene a Fe EERE FO ALLY NC ATT SE SE YN LT IE Et ~ a : are ba. rr) 9 234 MERCURY ARC VALVES shallow depression with a thin layer of mercury is provided so as to dissipate the heat efficiently. However, the evaporation of mercury vapour accounts for at the most 10 per cent of the total transfer, the rest being caused by vapour jets and splashing.» The principle of overcoming these phenomena has been known for a long time; this is to fix a cathode spot by some means and stop its crratic motion. There are two methods called anchoring, and film-emission, Tespectively, cach of which presents its own problems; although some progress has recently been made, and valves of small rating involving anchoring have been developed, much remains to be done before the work can be extended to high-voltage, high-current valves. In the absence of any direct method of controlling the cathode spot, it is necessary to provide some protection so that the vapour jets, which also contain large amounts of ionisation, do not reach the vicinity of the anode. Thus the anode structure (including anode, intermediate electrodes and grid) is situated in a long sleeve at the top of the tank with an umbrella-like baffle provided between the entrance of anode structure and cathode, the distance between anode and cathode being great. Sometimes two baffles are provided, one close to the anode structure and the other nearer to the cathode. The bafile arrange- ments may also incorporate oblique slots and fins to provide the necessary path for con- duction but completely block the direct straight line from cathode to anode structure; this system directs jets towards the walls and also provides extra surface area for deionisation. Unsymmetrical location of anodes (in multi-anode valves) in the arms connected to the central tank have been adopted by many manufacturers. Since the evaporated mercury should condense in the cooler, i.e., tank wall, regions of the valve it is important that it should not in any case condense in the anode structure, where condensed mercury could become a source of electrons when the anode voltage is negative and probably lead to backfire. This can be avoided by keeping anode structure at a higher temperature than the rest of the valve. After operating for some time at full load the anode requires, due to the anode losses, a sufficiently high temperature to prevent any nearby condensation and may, in fact, need cooling. At low loads or at the time of starting, however, it becomes necessary with high-voltage valves to provide some method of heating the anode and its surroundings; a separate anode temperature control will generally be necessary in high-voltage valves because of this. 13.9 Cooling arrangements There are several possible arrangements for valve cooling: (a) Forced air cooling. Forced air cooling generally consists of a fan placed directly underneath the valve. In this arrangement the air cools the cathode first, then the tank, and warm air exhausts past the anode structure thereby maintaining it at a temperature somewhat above that of the tank. This method of cooling is the easiest and simplest of all, although it is suitable only where the ambient air temperature is low. For large units, with a larger amount of heat to be dissipated, a large free area is required for sufficient exchange of air from the outside atmosphere and it may be desirable to provide extra blowers for this purpose. If the fan is not placed directly underneath the valve, the air is directed in the manner above. a - a eee " a ere a eens eA tl thal lk aL ir tov COOLING ARRANGEMENTS 235 In the ASEA valves for the Gotland Scheme, a fan for each valve blows air between the tank, which is provided with fins, and the outside casing. The anode structures (two in each valve, and which carry the anodes and intermediate electrodes) are surrounded by insulating sleeves and the air is forced between the two tubes by a separate air circulating system for cooling or heating of the anode structure as required. If the ambient air is impure, humid and consists of noxious gases, then the air cooling is done by circulation of pure air, which is cooled in a heat exchanger by water from an external supply. (6) Liquid cooling. Liquid cooling has the advantage of compactness and better control over the tank temperature. The most commonly used water cooling is direct water cooling, the water being circulated through jackets or pipes surrounding the tank and cathode. This is possible only if fairly pure water is plentiful. Water invariably contains impurities and in time is likely to corrode the inner surfaces of jackets and pipes. Because of this, water cooling is generally accomplished by recirculating pure water using a pure-water to raw- water heat-exchanger, or a pure-water to air heat-exchanger. Water cooling presents the problem of insulation. The tank being at a high voltage with respect to earth, the cooling system, which will be at the potential of the tank, has to Pas be insulated from earth. h In low-voltage valves, or valves in a circuit arrangement where the cathode is at earth the potential, which are provided with water cooling and a water-to-water heat-exchanger, this : problem is overcome by supplying raw water to the heat exchanger through a long insulating le pipe,“4) (42) so that there is no metal connection between earth and heat-exchanger (and * ¢ hence tank) and the long water connection through the insulating pipe provides sufficient uve resistance to any leakage. But in valves in which tanks are at a high potential with respect L: ; to earth, this insulation problem presents difficulties. It can be overcome by using a water- 3° to-air heat-exchanger, provided with radiators, and an air blower kept at tank potential, irby the power to.the blower being supplied through an insulating transformer. ‘y Water cooling also presents the problem of corrosion, particularly on account of the { gS : leakage current through water; to overcome this, the jacket, pipes, tank, etc., may be made Ais i of stainless steel, and a small percentage (about } per cent) of a corrosion inhibitor such as hydrogen sodium chromate can be added to the recirculating water.4) Diffusion of [: hydrogen is also likely to take place from water jackets into the tank. This is avoided by the provision of an impermeable layer on the water side of the cooling pipes or jacket.9) The anode cooling or heating must be by means of air. Oil cooling reduces the difficulties above. The circulating oil can be supplied through an insulating pipe and the heat exchanger (oil-to-air or oil-to-water) can be maintained at earth potential and, in fact, a common heat exchanger will suffice for all the valves. Figure 13.10 shows a line diagram of a suitable arrangement for liquid cooling of a group of valves,4) (42) The expansion tank is situated at a high level to keep the cooling system ' free from air and also serves as an expansion reservoir when the liquid becomes hot. A pressure-actuated relay provides protection against operation without liquid circulation. The exchanger (liquid-to-air) consists of parallel tubes provided with fins on the outside. The temperature regulator valve controls two valves; valve S, in series with the heat- exchanger, and valve P, in parallel with the heat-exchanger. The thermostat element 236 MERCURY ARC VALVES Lnsulolor Expansion lank pipes Pressure relay Temperature / regulator oo : valve Thermometer i Pum > Thermostat. circuit Liquid to air__. Q. =4-fan or heat exchanger i ees Figure 13.10 Liquid cooling arrangement for a group of mercury arc valves controls the positioning of the valves so that one closes while the other opens, thus regulating the proportion of liquid flowing through the heat-exchangers. Alternatively, the operation of the fan may be controlled, or the fan may be “on-off” controlled in addition by the maximum and minimum values of the temperature. A heater may also be installed inside the recirculating system to keep the temperature from dropping too low during a period of light load or starting. One important advantage of liquid cooling over air cooling is that a more even temperature inside the tank can be obtained; this is done in many multi-anode valves, by providing cooling pipes in the centre of the tank. In one case of a single-anode medium-voltage valve, cooling pipes are pro- vided inside the tank adjacent to the walls. Such cooling arrangements as these inside the tank also provide additional deionisation surfaces. om an IGNITION AND EXCITATION 237 Oil cooling has been adopted for Russian valves“) 426) in which provision has been made for circulating transformer oil in the jacket surrounding the tank and cathode pool. In the valve shown in Figure 13.2 the oil temperature is maintained between 14° C and 18° C. A novelty of the valves designed for the Donbass-Stalingrad h.y.d.c. transmission line,42®) is the way in which the hollow bafile is cooled by circulating oil through it, as shown in Figure 13.1 (@). 13.10 Ignition and excitation Ina multianode, multiphase valve the arc, once started, is picked up by one of the anodes and then transferred to other anodes in sequence by commutation. Such a valve conducts continually and hence ignition is only necessary at first starting. In single anode valves, however, the arc extinguishes every cycle. There are two ways of overcoming this problem; One by igniting every cycle at the desired firing angle and the other by adopting arrange- ments to maintain the arc once it has been started by the ignitor. The former are “ignitrons”, in which the cathode spot is present only during the period for which the anode conducts, and the latter are commonly known as “excitrons”’, in which the cathode spot is present continuously. With excitrons a grid is essential in order to prevent the anode from picking up current from the continuous ionisation above the cathode. In most ignitrons one or more grids are provided also in order to obtain rapid deionisation after the valve has stopped conducting and to obtain a higher voltage rating, e.g., a pentode ignitron, with two grids and one intermediate electrode, has been developed as a high voltage rectifier for use with radio transmitters.44) (143) Since the anode of the ignition arrangement has to be placed close to the cathode, it is gencrally desirable to provide an auxiliary anode, the excitor, near to the grid in order to pick up the current initiated by the ignitor and ensure quick and reliable firing of the main anode. Furthermore, since a free cathode spot is unstable for currents below about 5 A, this auxiliary anode is required to carry a minimum current of, say, 7 A to 10 A in order that the main anode current can be reduced to low values if required. With ignitrons the auxiliary excitation anode, if provided, conducts only for the period for which the main anode is conducting; with excitrons the auxiliary anode picks up the current from the arc initiated by the ignitor and maintains it continuously, the ignitor circuit thereafter being switched off. : Ignitrons have*been extensively developed in the United States for industrial purposes with the idea in mind that the provision of ionisation in the valve by continuous excitation during non-conducting periods is responsible for much arc-back phenomena (Section 13.15). There is no real evidence, however, that ignitrons are any more reliable than well-designed excitrons. In valves for h.v.d.c. requirements, it is essential to provide continuous excitation in order to ensure the reliable firing which is so desirable for smooth operation of the system and very important for the invertor. With the sharp current pulses that occur at the firing point of a high-voltage valve (Chapter 6) and which tend to quench the spot, the incidence of valve firing failure would increase considerably if the valve had to be ignited at every firing. 7 R 238 MERCURY ARC VALVES Continuous excitation can be of two types. In one type, the supply to the auxiliary are through the insulating transformer is first rectified by a dry type rectifier mounted on the valve chassis and then applied to the éxcitation anode. In the other type, two excitation anodes are provided and arranged to act as a full wave rectifier from the auxiliary single- phase supply, and hence maintain the excitation continuously. For single anode valves the former type is preferable, whilst the latter type is usually provided on multianode valves, ive., six excitor anodes for a six-anode six-phase valve. Ignitor and excitor arrangements are many and various. Three types, all of which depend on clectro-mechanical arrangements for breaking contact with the surface of the cathode and thereby striking an arc, are shown in Figure 13.11 (a),14) Figure 13.11 (6) and Figure 13.11 (c).949 Descriptions of these, and similar devices can be found in the literature. With ignitrons, a static type of ignitor is used“45) which consists of an ignitor rod made of refractory material of high resistivity, such as boron carbide, which is partly immersed in the mercury. To start the arc; a pulse of current is passed through the rod; the current through the rod heats the rod-mercury.junction and a film of vapour is produced around the rod separating the mercury fromit. An arc is thus formed, which is picked up by excitation anode; the ignitor rod then cools down in preparation for the next starting operation. The time taken for starting, if a steep-sided current pulse is used,.is of the order of 100 micro- seconds, and is known to cause no deterioration of the rod. Such a static ignitor is highly desirable for ignitrons if only for mechanical reasons, since the starting has to be carried out every cycle. Arrangements of this type have not been adopted for small excitrons since a pulse of 20 to 30 A at 200 V is needed for about 5; sec duration in order to give reliable firing and needs more complicated circuit than a mechanical type of ignitor. For h.v.d.c. purposes it is very important to re-ignite the arc as rapidly as possible after excita- tion failure and it is the practice of ASEA of Sweden to use ignitors of this type; a capacitor is discharged through the ignitor and produces enough energy for ignition. 13.11 Gaseous impurities and pressure control A valve structure has several joints. The anode structure will generally consist of insulator rings (glass or porcelain) separated by, and scaled to, intermediate electrodes, the anode will have its lead brought out through the top cap, and there will be an insulating barrier between the tank and the cathode. There will also be ignitor and excitor leads to be brought out through insulators sct in the metal of the tank. Such connections will require efficient seals of cither clectrode-insulator (graphite) type or insulator-tank (steel) type. Furthermore, the tank itself generally consists of a cylinder with top and bottom plates joined to it to provide reduced diameters for connecting the anode and cathode structures; these tank joints involve metal-to-metal seals. The above joints may be permanent (fused and welded) or demountable types, ¢.g., mercury scals. A high-power valve will generally consist of both in order to facilitate dismantling of the valve for repair and servicing. The joints, particularly the demountable ones, are always likely to permit the ingress of air, which infiltrates also through the walls due to porosity. The air will increase the pressure of the valve in spite of efficient tempera- ture control and will thereby reduce its breakdown voltage. It has been suggested that for omen caee e ne). kee Sleel plunger wilh cup-shaped lop containing mercury. (Position 239 when soknold Secler | : unenergised), XC“ | So opens when excifer ~ h picks up current and ¢ < ode poo! < energises a | the relay coil C M The sokenold and igniter are in series and the plunger 1s pulled down fo slart an arc which is then picked up Solenoid aC. supply by the excitation electrode; Ha's operation causes the relay So lo be energised, thereby cutting off the igniter circuit (2) Solenoid ecier S2 opens when exciter picks up current and 3 / energises / relay coil. ¢ S) < a SQ ie S; closed when starting Cathode pool Insulated plunger (position when lec. solenoid unenergised). Supply The solenoid and ignilor are connected in parallel The solenoid pulls the igniter downwards to make confact with the mercury surface; this short- circuits the solenoid and the ignitor i's released fo start an arc. ©) Figure 13.11 Examples of ignitor and exciter arrangements sie. 240 MERCURY ARC VALVES U a Exciter Jet of mercur = == === Cathode pool —___ Solenord Annular iron Plunger When the solenoid 1s energised, the plunger Is pulled downwards lo cause a jel of mercury fo strike the ignitor, which Hen starts an arc A arcuit arrangement similar to that of (0) 1's provided. (C) Figure 13.11 (c) Example of ignitor and exciter arrangements valves up to 1,000 V the partial pressure of impurities should not excced 0.1 micron, at least in the anode neighbourhood, whilst for high-voltage valves, i.e., 10 kV or more, the impurities should be kept at about one-tenth of that value.®® Leakage of air is not the only reason for loss of vacuum. Gases are absorbed by all materials to some degree, and are occluded when heated. Gases are removed as far as possible by first raising the temperature of the individual parts as much as possible and then operating the assembled valve for a considerable time on a low voltage at higher than rated current in order to raise the temperature above that for normal rating, the whole assembly being continuously evacuated. The assembled valve may have severe degassing tempera- ture limitations, due to the different temperature expansion coefficients of the various materials used in its construction. Mercury is likely to contain impurities which may be given out gradually during operation. The mercury is given a cleansing treatment which consists of working-in nitric acid and dis- tilled water; mercury which is suspected of containing oil is given an extra washing with either gram alcohol or chemically pure benzol.@40) During assembly a certain amount of forcign material is unavoidably left in the valve, such as lint, grease, dust, etc. In order to remove contamination, a process called ‘“‘aera- tion” has been used which consists of cooling the rectifier and admitting clean dry air fora few hours. This air is admitted through a tube containing a filter and a drying agent such as activated alumina. The air tends to oxidise the contamination and is pumped out prior to degassing. 40) = pe prem py a = 3 e ek BoA PUMPING ARRANGEMENTS 241 Mercury arc Vac. valve Hot wire voc. valve gauge Mercury vapour Vac. pump VacSmeler Sn 8 Mcleod vac. i a Fi Aufomatic vac. valve Rolary decgell Figure 13.12 Pumping arrangement for a mercury arc valve The thermal gradient, falling from the anode towards the cathode, can be used with effect; thermal diffusion tends to drive the impurities from the hotter anode region to the cooler cathode region, where they can be removed by the pumping arrangements. At present pumpless valves are only provided for certain industrial installations where the cost of pumping equipment is relatively high, and it is found to be preferable to replace the valve in case of loss of vacuum. In high-voltage valves a pumping system is inevitable, on account of the large number of joints, the necessity of providing some demountable joints, the need for a low operating pressure, inadequate reliability of joints at such low pressures, and the low relative cost of pumping arrangement compared to the total cost. 13.12 Pumping arrangements The commonly used pumping or evacuating unit for mercury vapour valves is not markedly dissimilar from that used for other purposes and consists of two vacuum pumps in tandem, a mercury-vapour diffusion pump and an oil-sealed rotary pump, as shown in Figure 13.12.94) The rotary pump generally provides a backing pressure of about 10-2 mm of mercury for the diffusion pump, which in turn provides the required pressure in the tank. The tank is connected to the pumping unit through a vacuum valve which is closed when the pumping unit is not in operation. A vacuum valve.between the two pumps closes auto- matically when the rotary pump stops, in order to prevent oil from leaking into mercury vapour pump. If the valves are at a common cathode potential in a converter, a common pumping set is generally provided. In a bridge circuit, however, a separate pumping set is required for wa 242 MERCURY ARC VALVES each valve; economy in the pumping system may be obtaincd by providing the mercury vapour pumps with a pre-vacuum tank instead of pre-vacuum rotary pumps.® In this case, only one rotary pump will be necessary to evacuate the pre-vacuum tanks of each valve after long intervals. For vacuum measurement one or both of two types of gauges are generally in use: (a) McLeod vacuum gauge. (6) Hot-wire Pirani type gauge. The difference between the compression vacuum gauge (McLeod) and the hot-wire gauge is that the former indicates only the pressure of non-condensable residual gascs whilst the latter measures the pressure of both non-condensable gascs and vapours. Some pressure indication will also be necessary on the pre-vacuum tank, so that it can be maintained at an adequate state of evacuation; a simple mercury manometer is adequate for this purpose. The pumping equipment, like the cooling equipment, is connected to the valve tank and has to be insulated from ground. Also it should not have any metallic connection with the cathode and equipment connected to it. The supply to the diffusion pump heater, hot-wire vacuum gauge, etc., as with the other auxiliarics, is through the insulating transformer. 13.13 Ion bombardment and electrode sputtering . When a valve ceases to conduct, the anode and other electrodes suddenly acquire negative voltages and there is a heavy bombardment of their surfaces by positive ions. This bombardment continues, though to a lesser extent, for the rest of the negative part of the applied voltage due to infiltration of ions through the grid, which is covered by a sheath of ions. The ions lose their kinetic energy on impact to produce X-rays, heat, displacement of clectrode material (called ‘‘electrode sputtering”), and electron emission. There are a number of theories of electrode sputtering, one of which suggests that the energy of the impinging ions raises an area around the point of impact to a temperature which is instantancously very high but which sinks rapidly as the energy spreads out and is dissipated by the thermal conductivity of metal particles. Another theory suggests that ion bombardment fractures the surface crystalline layer, and the sub-microscopic dust thus produced is quickly evaporated when it is passed into the discharge by the heat of recom- bination of ion and electrons on its surface. The sputtered material consists of fine dust moving away from the electrodes at various speeds. Their movement is influenced by gravitational force, electric fields and further collisions. They finally settle down on various obstacles in their path, piling up in flake-like lumps on electrode surfaces, insulators, walls, ete. It is possible for sputtered material to bridge the inter-electrode gaps and cause short-circuits, or cause microscopic sparking which produces, in its turn, extra ionisation and reduces the voltage withstanding capacity of the gaps.89 The metallisation of insulator surfaces by sputtered dust can impair the insulation, and it is even possible that the grid could receive its potential from the anode instead of from its own source, and consequently lose its control ability. The amount of sputtering depends upon the type of electrode material and the gas, and increases with voltage, current, ionisation and temperature. Graphite is frequently used for electrodes because it has a low sputtering rate and good degassing properties. To eee “ssure agin ss v and hpre VALVE RATING AND ARCING FAULTS 243 resist sputtering, the material must be hard, with crystals strongly bound together, and at the same time possess porosity for efficient degassing. On degassing, porous graphite readily gives out all its gaseous content and, when the valve cools down after degassing, it absorbs any remaining traces of gas and frees the valve space from foreign material. This property is not, however, so important in a valve provided with continuous pumping. In high-voltage valves, in which the intermediate electrode design enables them to act as obstacles to the charged particles, it is most impor- tant to use material for the electrodes which is extremely resistant to sputtering. 13.14 Ion deficiency and spot quenching In mercury-are valves high current conduction with a low voltage drop takes place because of the vapour ionisation by which positive ions neutralise the electronic charges in their path. This ionisation increases with increase in current. If, at a constant vapour density, the current is further increased, a stage will be reached at which all the vapour will be ionised. An increase in current beyond this point will result in a condition where the number of ions are insufficient and the valve will behave in a manner similar to a high- vacuum valve with increase in voltage drop across it. In fact, the voltage may start rising earlier than the saturation point.447) ‘ It has been demonstrated that this phenomenon results in cathode spot extinction and current interruption; the abrupt current interruption producing high-voltage surges in inductive and capacitive circuits. This condition limits the current density of the valve under short-circuit conditions, and is fixed by the vapour density (and hence the current) in the narrowest cross section of the discharge path. This critical current may be about 5 A/cm? for an operating temperature of 20° C, rising linearly to 60 A/cm? for 50° C.S9) In practice, the valve is operated far below the critical ionisation and this point is only reached during heavy overloads under extreme operating conditions. But if the current interruption does not occur, it is very likely that permanent damage could be caused to the valve because of increased losses and overheating in the event of persistent fault current. Spot quenching is more likely to occur from a current surge with a high rate of rise. These surges can occur during the firing of a high-voltage valve when the existing spot corresponds to about 10 A excitation current. The rapid rise of current produces a faster rate of clec- tron removal than the maximum rate at which the spot emission can expand, and thus there is exhaustion of the spot encrgy;®® this exhaustion of energy causes stoppage of the emis- sion, with consequent high-voltage stresses. 13.15 Valve rating, arcing-back and arcing-through phenomena In a valve with intermediate electrodes, an arc-back starts with a breakdown of one gap, which produces an increased stress on the other gaps together with some diffusion of ionisation from the broken-down gap into the adjacent sound gaps; this causes breakdown of other gaps and finally of the whole valve. Positive breakdown (arc-through, failure of grid blocking) will occur in a similar manner with the breakdown of one gap followed by others until the stress between first electrode (next to the grid) and the cathode will be such that the grid will be unable to stop the diffusion of electrons. It follows that if the gaps are (dd sad te ne la en ts 244 MERCURY ARC VALVES rated to withstand safely the voltage which could occur in the case of breakdown of one or two gaps, the valve breakdowns would be very much reduced. Various investigations have been made into valve failures and their causes, and these investigations have led to increased valve ratings and a great reduction in the incidence of breakdowns, but not complete climination. The causes are numerous and a small per- centage of breakdowns appear to occur at random, at any time of the cycle and at any load. It was pointed out in Section 13.2, that the breakdown voltage of a gap depends upon the gap length, vapour density, electrode material and residual ionisation. With the com- mencement of commutation of current from one valve to the other, the current in one valve starts to decrease and so does its ionisation. The instantaneous rate of deionisation depends upon the amount of ionisation present at that instant; thus, as the current decreases, the ionisation decreases but the rate of its decay decreases, and as the commutation proceeds, deionisation will not be able to keep pace with the rate of current decrease, there being some residual ionisation when the current zero is reached.449) y An increase in the incidence of breakdowns can thus be expected with increase in initial \ voltage jump [instantaneous phase to phase voltage at an angle (a+y)], current and applied voltage. Furthermore, with the above factors constant the breakdown will tend to increase with decrease in the magnitude of the commutating reactance, since this factor will increase ‘the rate of change of current during commutation. Thus the valve performance does not j depend upon its constructional features alone but also upon the system and transformer (reactances and the nature of its connections. Most tests indicate that with grid delay, the greatest number of arc-backs occur at the initial voltage jump where the residual ionisation is a maximum. Following one test series, it was suggested that the rate of occurrence of arc-backs was proportional to the product of the initial voltage jump and the rate of change of current at the end of com- mutation. This is quite compatible with the discussion above, since with increasing angle of delay, both the initial voltage change and the rate of change of current increase, and a corresponding increase in the incidence of arc-backs is likely to arise.449) The rate of occurrence of arc-backs at the initial voltage change will be reduced by the filter circuits, which are provided to reduce the parasitic oscillations accompanied by the sudden change in voltage. These filter circuits will decrease the slope of the leading edge of the voltage waveshape.8) Although a large number of breakdowns can be attributed to the initial voltage change, breakdown occurs also at subsequent instants. It can be seen from the inverse voltage’ diagram of the rectifier valve (Figure 3.18) that the inverse voltage continues to rise after the initial voltage change accompanying commutation. The ionisation continues to decay and hence the withstanding voltage capacity of the gaps continues to rise. If the inverse voltage is i the increase in withstanding voltage capacity, breakdown could occur; such a breakdownis likely between the initial voliage change and the maximum voltage peak.s8) ! In inverter operation, breakdown at a negative-going voltage change is Icss probable, since it is generally small, unless the angle 5 happens to be large (Figure 3.12). After this sudden change the voltage decreases and changes sign. Before the voltage zero, the dicak. VALVE RATING AND ARCING FAULTS 245 deionisation should decrease sufficiently to enable the grid to regain control and break the continuity of ionisation. Thereafter, the breakdown can occur if the positive voltage exceeds the breakdown voltage of a gap or if it exceeds the voltage the grid can control. The ionisation present in the anode structure during the non-conducting periods is not only the residual ionisation from the decaying ionisation of the preceding current. Some of the ionisation from the main body of the tank, where a small amount of ionisation is main- tained by the excitation current, will infiltrate through the grid and subsequent electrodes. This infiltration can be caused by the available vapour expanding during conducting periods and contracting during non-conducting periods. This effect can be aggravated by the existence of random vapour jets which can bring a blast of vapour at high pressure near to the anode structure entrance. Although the bafile system will prevent the jets from coming near to the actual anode structure, their secondary effects are likely to be noticeable. Pressure transients can be caused by mercury droplets striking any hot objects after being splashed from the cathode spot. These droplets, travelling at relatively slow speeds, may well reach the upper regions during the non-conducting periods long after they have been created by the full current; this may account for some of the random breakdowns.49) It has also been observed that valve failures increase after a valve has been out of ser- vice for some time; this is mainly attributed to the mercury droplets which have condensed in the anode structure when the valve was allowed to cool down, and can be avoided by heating the anode structure prior to starting up again. As mentioned above, good tem- perature control is essential and it is necessary to keep the anode structure continuously at a higher temperature than the tank. Impurities of various kinds are generally considered to be responsible for most of the breakdowns, and for their random nature. One theory suggests'150 that the arc-backs are basically due to positive ions accumulating on microscopic insulating particles of impurities, which are then carried on the electrode surfaces by the vapour streams. If the charge is sufficient for the potential gradient to exceed the field emission requirements of the base material, breakdown occurs. Whether or not sufficient potential gradient is established depends upon the rate at which ions strike the insulating particle and the rate at which the charge leaks away; the latter depends upon the size and leakage resistance of the insulating patch. The rate at which the ions strike the negative electrode is proportional to the ion density in the space in front of the electrode at the end of the conduction and the negative voltage. By this theory, the observed increase in breakdowns with increase in initial voltage change and rate of change of current can be expected and it also accounts for some random breakdowns. The fact that impuritics in the electrodes may be playing an important part is supported by the fact that when a voltage applied across a gap is raised, incipient breakdown is indicated by the appearance of scintillations,03) “5 or bright spots on the electrode surface. These scintillations would result in full breakdown if the current was not limited. If the voltage was applied by means of a capacitor, so that a breakdown of such a nature would result-in a pulse of current only, after a few repetitions, the tendency would be for no further scintillations to occur at that particular spot on the electrode at that particular voltage; as the voltage was increased, the scintillations would reappear. It has been suggested that this may be because of the destruction of the impurities which caused the ak Git 246 MERCURY ARC VALVES original scintillations, and that with an increase in voltage, the impurities in a deeper layer are reached by the bombardment of ions with a higher velocity. This phenomenon, apart from the degassing treatment mentioned in Section 13.11, leads to a valve also being given pre-service treatment under conditions of low current but with gradually increasing voltage to a level higher than the normal working voltage. During normal working, vapour and gascous impuritics may also be trapped in the clectrodes, especially in a deposited film of sputtered material, and there may be occlusion of gases from inner layers to outer layers under conditions of heating. It is not unrcason- able to postulate that the random breakdowns, given a correct assessment of valve rating, are almost entirely dependent on impurities in electrode material, mercury, etc., and not on valve parameters and operating conditions. Apart from sparking and scintillations on the electrode surfaces there is a tendency to sparking at the electrode-insulator boundary which tends to cause breakdown, (103) (152) Unlike the scintillation phenomena, it is likely to reoccur in the same place at a lower voltage. The reason for this is not very well known and this may not be true forall insulator materials. Good design arrangements to minimise this kind of breakdown necessitates keeping the electrode-insulator boundary somewhat away from (preferably placed in deep recesses) a direct path of ion bombardment. ; The voltage withstanding capacity of a valve can be increased by decreasing the ionisation density of the vapour. Thus it has been observed that a valve operating at a current lower than its rated current can withstand voltages higher than its rated voltage and vice versa. In fact, within certain limits of current and voltage, a valve may have a constant kVA rating.(48) To fix the permissible voltage rating of a valve, account has also to be taken of the over- load currents, of about 1.5 to 2 times, which the valve may be required to carry without losing control or incurring breakdown. In high-voltage valves the normal working current density may be of the order of | to 1.5 A/cm2, and the area of cross section to be considered will be the free vapour passage available through the electrodes in the anode structure.(30 Due to the great interdependence of the voltage rating and the current rating, it is impor- tant that the working current never greatly exceeds the rated full load current. On the other hand the valve has a fairly high short-time (up to, say, one or two cycles) current rating and can safely carry the short-circuit currents that occue during backfires, com- mutation failures, etc., provided the valve is allowed time immediately afterwards for deionisation and recovery before starting its normal operation again. Fortunately, satisfactory grid control arrangements and provision for blocking provide complete safety for the valves (Chapter 6); also, with the deionisation or recovery time being small, a valve can be put back into operation after a very short time. For the Channel project, for instance, the valve group which has been blocked will be started again after 0.2 sce; _there will thus be no interruption in transmitted power. Although it may be expected that the current capacity of a valve can be increased by increasing its size and maintaining the current density and the vapour density within reasonable limits, the reality is not so simple. With valves of small current ratings, the deionisation is rapid and equilibrium between ionisation and anode current is reached quickly so that the ionisation density is relatively low when inverse voltage is applied at ee coer eet eae RR 2 6 alah tl hina lila aman se } TANK POTENTIAL WITH RESPECT TO CATHODE 247 the end of the commutation period. As the cross-section and volume of the valve increase for higher current ratings, the volume of ionised gas increases and the time required for ions to diffuse to the walls is increased with the result that the density of ionisation at the end of the conduction period is correspondingly higher. This not only results in an in- f a i crease in dcionisation time but also reduces the voltage rating of the valve. It has thus [.. i been observed that time required for deionisation is greater for larger valves of similar sion | construction, for the same current. To overcome this type of difliculty means must be | adopted for rapid deionisation, and as discussed in Section 13.3, these can be provided by ‘e ample surface area of grids, bafiles, insulators, electrodes, etc. It appears that, with these difficulties overcome, valves of practically any current rating can be designed. In the valves of the Gotland scheme, two anode structures, each with their own grids and inter- | mediate electrodes, have been provided) ) and the valves designed by the same firm : (ASEA of Sweden) for the English Channel project have four anode structures in parallel.'52 This method may have been found more suitable and economical compared with increasing the current rating by enlarging the anode structure. ya 13.16 Tank potential with respect to cathode"l4®) dy ' It is undesirable to connect the tank directly to the cathode, because of the possibility = of the cathode spot forming where mercury has condensed on the tank walls. Such T occurrences, although theoretically inhibited (due to the length of discharge being smaller than the critical length § and hence having a tendency to take the longest path (Section 13.2.1), have been experienced and would clearly erode the tank and damage the valve. An electrically isolated tank, together with the auxiliaries metallically connected to it, will have a considerable capacitance with respect to earth. The tank is electrically con- nected to the cathode, however, through the medium of ionised vapour, and its potential will follow the cathode potential with respect to earth. When the cathode potential goes bi} Go negative, current flows from tank to cathode through collection of electrons by the tank, = fe which by this means accommodates itself to the cathode potential. ; (: When the cathode goes positive, the current flows through the collection of positive Tent ions by the tank; because of the small mobility of positive ions and their limited number, ale the current in this direction will be limited and the tank will not be able to adjust itself to fr i the change of cathode potential in the positive sense. As a result the tank may acquire a high negative potential with respect to the cathode, and may act as a crude negative grid viele i to become a source of extinction of the excitation arc. This possibility can be eliminated E: by connecting a low-voltage capacitor of about 1“F (much larger than the tank-carth tie capacitance) or a non-linear resistance between the tank and cathode; this will serve to sain | maintain the tank potential very nearly to that of the cathode. [ | the 13.17 Anchoring and film emission ! shin Mention of experimental work on anchoring and film emission is made here since they L may well be of importance in future developments in high voltage mercury valves. The ' 7 idea of anchoring comes from the tendency of the cathode spot to fix itself in position under the special conditions arising at the junction of the mercury and some other metal protruding [ 248 MERCURY ARC VALVES out of it; the spot becomes a bright line whose length depends upon the current. This was noticed independently by Hewitt'5? and Weintraub"55) and later developed by many other workers. Insufficient has been known about anchoring and no significant com- mercial use of it has been made until recently. The phenomena of mercury jets and splashing do not take place after the spot has anchored itself. Anchoring is successful only if the anchor has been wetted by the mercury, and is thus possible only with the metals which spontancously form amalgams with mer- cury. Wetting is the formation of a concave meniscus at the contact surface between mercury and metal instead of a convex meniscus formed without wetting, as illustrated in Figure 13.13 (@). Wetting action has to be achieved only once, thereafter an interface is Wetted line Cathode lead-in molybdenum Cylindrical molybdenum or tungsten rod protruding anchor piece pressed on from mercury surfoce lead-in rod Strip anchor(molybdenum) — Group of parallel anchoring supported by lead-in rod. slrips(molkybdenum) pressed dn steel block, vibich itself (s pressed ono lead-in rod. Molybdenum strip, coil fo spiral and riveted to lead-in rod in centre. Figure 13.13 Experimental anchoring systems from 1930 to 1940 Ee reece re ei Ee es 2) tO Sanaa aii ae ANCHORING AND FILM EMISSION 249 formed at what was previously the contact surface of metal and mercury, anchoring then being obtained every time. A phenomenon has been observed in the process of obtaining the initial wetting. The formation must be started with a small free spot; this free spot, moving erratically, approaches the anchor occasionally and in one of these approaches, wetting will occur, the spot will anchor itself and the wetting will subsequently extend along the mercury-metal surface as the current is increased.(92) (94) (99) Various forms of anchors have been tried?) 9 in the form of rods, plates, spirals, etc., as shown by Figure 13.13. With all these anchors it has been noticed that the anchoring ceases after the current intensity of the anchor line is increased beyond a certain limit, this limit being different for different anchors: this was demonstrated in particular: by Amillac with molybdenum spirals floating in the mercury pool, there being an increase in length of the emission line with current up to a limit of several hundred amperes.(156) Another point of importance’) is that if more and more anchors are provided, the ten- dency to anchor decreases as soon as the spacing between anchors becomes less than 1.25 cm, approximately. To increase the current limit by increasing the anchor line, and hence the cathode size, is technically difficult and an economically unrealistic approach since the cathode would then have to be several times larger than the normal cathode. A subject of much investigation has been the search for suitable materials for anchors which would not erode rapidly, and would also have a very high melting point. So far tungsten and molybdenum have been found to be suitable, but both of them are very difficult materials for manufacturing the required cathode sizes and shapes. Molybdenum is preferable, however, although its preparation is difficult because of brittleness and the tendency to crack which would prohibit the maintenance of a vacuum.2) Anchoring has been introduced into the design of industrial valves, although this has been mainly to achieve stable operation by anchoring at low currents.(150) Von Bertcle and Steenbeck®) carried out a series of experiments on various types of anchors which indicated that the success of anchoring depends upon the heat dissipation and that, by improving this factor, the maximum current at the anchor line could be in- creased. Hence it was found necessary that the thermal conductivity of the archor line should be made very much higher than the thermal conductivity during the condivions of a free cathode spot. ! Figure 13.14 shows the two types of water-cooled anchors developed for experimental purposes. The first one, Figure 13.14 (a), is a 10 mm drilled molybdenum rod which successfully carried a current of about 125 A at 17° C inlet water temperature; the second type of anchoring shown in Figure 13.14 (6), and involving a molybdenum cup mounted on a cylindrically-shaped iron cooling system, was developed to overcome the limitations of molybdenum rods. It was noticed that anchoring action was obtained at some places only which, on examina- : tion, showed to be due to the discontinuities of the joint between the steel and the molyb- denum. Thus it was shown that there should be a perfect molccularly-undisturbed bond (similar to that obtained in welding, soldering, brazing, etc.) between the molybdenum and the outer metal and that if the heat conductivity along the anchor is not uniform, the spot has a tendency to concentrate at particular points where the heat conductivity is on Pe ‘ V ago vl — mn rs nner eee ene eatin pe ne A CE a 250 MERCURY ARC VALVES greatest. This led to a very import ant conclusion th emission at temperatures much lower than that obt Various other features observed with anchoring (i) The mercury at the wetting surface has a smaller work function than the pure mer- cury, thus involving a reduction in cathode voltage drop of about 1.5 V. (ii) Most of the cathode losses are dissipated at the cathode, and thus little or no mercury transfer takes place. 7 (iii) The length of the emission lin the vapour pressure. As the along the wetted line until th the increase of length of emissi at successful anchoring involved ained at the free spot. were as follows: ¢ increases with current but primarily depends upon vapour pressure is increased the anchor line extends € maximum length is covered. It thus appears that on line with growing current appears to be a secondary feature associated with an increase in local vapour pressure. (iv) The current on an anchor can be maintai compared to at least 3 A in a free spot. These experiments, continued in England by Von Bertele ment of a commercial 50 A, 2 kV (maximum values) valve ned to a value as low as about 0.1 A after 1946, led to the develop- with air-cooled anchoring, the Netting line Molybdenum lubular anchor made Molybdenum cup from solid on hallow molybenum rad steel carrier Figure 13.14 Experimental low-heat-resistance anchors ANCHORING AND FILM EMISSION 251 cathode arrangement of which is shown in Figure 13.15 and consists of a molybdenum cup containing mercury.©2 Gov The most important fact arising from this work is that the transfer of mercury is a matter involving the valve atmosphere and that by keeping the emission surface temperature lower than the temperature of the vapour above, the mercury transfer from the pool to the chamber stops and, in fact, reverses; i.e., the mercury condenses on the cooled surfaces of the cathode with what is called negative evaporation as predicted, in 1928, by Issendorff.057) Furthermore, the variation of the temperature difference between the cathode and the chamber revealed different phases of emission, most important among which is the spreading of the steady emission area over the mercury surface and the film of mercury condensed on the cool surfaces of the anchor; this phase is called “film emission” or “patch emission’. This phenomenon was also observed by A. J. Humphrey® when the heat dissipation from the discharge-vessel walls was greatly suppressed and that through the mercury pool was increased by fitting a metal bottom with cooling fins under a mercury pool only 5mm deep. With such a system, the current density of the emission area decreased with in- creased wall temperature and could be reduced to approximately 60 A/cm? with emission currents as high as 150 A per spot. A smooth mercury surface without local depressions at the emission area indicated almost complete cessation of the vapour blasts. Kobel,“58) in 1930, reported an evaporation as low as 0.017 mgm/A sec compared to 5 mgm/A sec with a free cathode spot by providing only a small mercury surface area of 1 cm? and reported that the emission spread uniformly over the surface with a current of 35 A. Molybden um cup Aluminium fins —--- eS Tin solder |) Nickel plating // Mercury / vA L Figure 13.15 Air and heat flow lines around a low-heat-resistance, molybdenum cup cathode asad Pine ode ey 252 MERCURY ARC VALVES Ww v | / S 30 wy! 6 | | S of Le) 3 ate ol | QV solLe! $! &Glowless | : & yet v) Oy ; 7 Figure 13.16 Zones of OD W; iw ev enmn1sslon | preference for the individual ~ | YX / | emission phases © INIG/ | | v 4 —h & 9 if 1 | [Fila emission, S 1) with glow | ® gored ft ---f -=-p 1} 8 la, 9 ‘50 /00 200 300 400 C Wall temperature To study the different phases of emission Von Bertele has carried out experiments at Imperial College, London{5® (60) with a special system consisting of a molybdenum emission surface of about 1 cm?, with provision for controlling the temperature between 20° C and 70° C; the whole was enclosed in a discharge vessel with several electrodes and, below the molybdenum emission surface, a reservoir filled with about 1.5 cm3 of mercury was provided. The walls of the discharge vessel, except for the cathode surface, were maintained at a uniform temperature which could be varied between 20° C and 350° C. The flow of vapour towards the cathode system eould thus be controlled and determined by the temperatures of the cathode and the vessel. A protective conical screen of thin sheet extended from the temperature-controlled surface into the mercury reservoir and prevented damage of the glass seal by the discharge. The discharge current covered the range of 1 to 30A. Over the varying range of temperatures, five different phases of emission ap- peared with increasing temperature difference between the vessel and emission surface, as follows: (i) The free spot, moving randomly on the top of the reservoir of mercury and on drops of mercury condensed on the protective cone. (ii) Transitional stages, consisting of spots and fragments of discharge lines moving at random but with a preference for the temperature controlled surface. om lL [ Pec yp ANCHORING AND FILM EMISSION 253 (iii) Streamers, apparently consisting of a multitude of very small, quickly-changing spot elements. (iv) A uniform and glowing sheath, of very small height and little brightness covering, and clearly restricted to, the temperature-controlled surface. (v) At high metal temperatures and the highest currents available, a disappearance of the glow-like sheath with smooth transition of the current into the discharge. The various phases change over from one to the other as indicated in Figure 13.16 and, at low temperatures, could overlap. These experiments are not yet completed. In all the phases above the only common feature is the low are drop, with very little change throughout, indicating that one and the same process is responsible for the electron emission. Film emission, to the exclusion of any other form of emission, can be obtained over a wide range of current densities: although at very low values, or with insufficient vapour density at the emission surface, the even spreading ceases and contraction into lines, streamers or spots takes place. It appears that emission from the surface of liquid metals docs not require high current densities of the order of thousands or tens of thousands of A/sq cm, nor does it require evaporation of metal. Thus the emission theories which are mostly based on high current densities and evaporation of cathode material cannot claim universal validity. The main advantage of a film emission cathode over an anchored cathode is clear: the uniform spreading of the emission over large areas permits the handling of large outputs of 600 A or even 1,000 A per valve, whilst a valve with anchored cathode may have to be restricted to 200 A. Both systems, however, have advantages over free spot emission in that evaporation and vapour jets are eliminated; a large condensing tank is thus unneces- sary, and the electrodes can be brought nearer to the cathode thereby producing a reduced voltage drop, elimination of baffles, steadier performance, and improved grid control. The main difficulties with the application of film cathodes to high voltage valves will undoubtedly be on account of their low operating temperatures and vapour densities, and hence very limited range of allowable temperature fluctuations in order to obtain consistent film emission. me Cf oe oe os CHAPTER 14 Direct Current Cables References: (2) (9) (17) (23) (53) (72) (74) (126) (180) to (208) inclusive 14.1 Introduction Cables for direct current are not new and were brought into service in the early years of the century in both France and England; these cables were of the impregnated paper or “solid” type, and are the only type which so far have been placed in industrial service, Cables based on polythene-insulation are now a strong possibility and the subject of exten- sive research, 480 Distinct differences exist between a.c. and d.c. insulation phenomena, and have a marked effect on the approach to cable design in each case: (i) In the a.c. cable case, the radial dielectric stress distribution is dependent upon the permittivity of the dielectric, where in the case of d.c., the stress distribution is determined by insulation resistance. The permittivity of a dielectric is only very slightly affected by a variation of temperature over the working range and hence very little change of stress occurs because of it; the insulation resistance of the dielectric is extremely temperature-dependent, however, and gives rise to appreciable variations in stress distribution as a result of change of temperature in the d.c. case. (ii) In the case of an a.c. cable, the maximum stress always appears at the conductor surface, but with d.c. the maximum stress may appear at the conductor surface or at the outer boundary of the diclectric, depending on the temperature or temperature gradient conditions. (iii) The d.c. strength of a dielectric is much higher than the a.c. strength and in con- sequence much higher diclectric stresses can be used for the former than for the latter. (iv) With a.c. cables the temperature limitations are set by the physical behaviour of the materials and the method of construction, whereas with d.c. cables the temperature limitations are set not only by the physical limitations of the material and the struc- ture but also by variations in dielectric stressing arising from temperature effects. 14.2 Basie physical phenomena arising in d.c. insulatioa{sD (3) Several different factors limit the working stress with solid insulation. There is the intrinsic electric strength, which is never achieved in practice, and which has a very high value of the order of millions of volts per centimetre. For impregnated cables under pressure the effective intrinsic strength is not known-with any accuracy, but is certaialy in excess of 2 million volts percm. There is also a maximum thermal voltage, independent of thickness of the diclectric, which depends on frequency and on the relation between the dielectric losses and the thermal resistivity of the insulation; for good industeial insulation rn ot me ae ee i ooeeeheeneannamemntenmenmemmanenementn ne [ BASIC PHENOMENA OF D.C. INSULATION 255 Figure 14.1 Electrical model for dielectric breakdown Co r this is usually somewhat below 5.10° V, r.m.s., at 50 c/s, and which clearly does not set a limit for the foreseeable future. The limit in practice with a.c. is determined by the onset of discharges in voids in the geo dielectric. In Figure 14.1, C, represents the capacitance of a gas-filled hole in a solid |: dielectric, C, that of the solid dielectric which is in series with it, and C, that of the mass of ToC. dielectric which is in parallel with the C-—C, combination. The gas may have a lower permit- Nten- ; tivity than the mean permittivity of the dielectric, and thus its voltage gradient will be higher, [ the increase depending on the dimensions of the hole. If now with voltage V kV across the bed whole system, the gas in the hole breaks down, the voltage across C, will fall very rapidly: the discharge will then extinguish, but during the discharge energy will be released in the 7 i void, with the possibility of some damage. aps If the voltage V across the whole system is further increased, there will be a new discharge wry ' for each step of V volts, and similarly with each step of V volts when the voltage is reduced. ay In consequence, with a.c., if the insulation is stressed above the “discharge inception i voltage”, discharges will occur each half cycle and the number of discharges in a given void een will depend on the total change in voltage. Whilst solid insulation has worked and does work satisfactorily for years with a few discharges, it is not safe to work it much above the fT ; discharge inception voltage, since with rise in voltage more and more voids break down and gat the frequency of recurrence of the discharges in each void increases. ati With d.c., each void will also break down at each multiple of its inception voltage as the voltage rises, but once the maximum voltage has been reached, the capacitance can only be recharged by leakage current. Discharges thus recur only at intervals dependent on the time constant of the insulation, which may be minutes, hours, or days. ‘ If, therefore, the insulation supports an alternating voltage V kV, r.m.s., determined by the onset of discharges, then since the number of discharges is determined by the total voltage change, the same insulation might be expected to support a direct voltage of at least 2 V2VkvV. Further increase with d.c. is a possibility because of the lower repetition frequency of individual discharges. With mass impregnated cables, the normal working stress at 50 c/s is about 70 kV (peak/cm); the corresponding total voltage change with a.c. is 140 kV, so that the permissible d.c. stress may well be of the order of 150 kV/cm. With modern supertension cables, in which the incidence of discharges is greatly reduced by the use of a filling medium of gas or oil under pressure, normal! working stresses at 50 c/s lyin : may be 100 kV (r.m.s.)/cm: the corresponding stress with d.c. working would be 300 kV/cm, { or about one-fifth of the electric strength which can be attained by impregnated paper cables iM under 300 Ib/sq in. pressure. If over-voltages two or three times the working voltage were to be allowed, which is very unlikely in h.v.d.c. system, then conditions in service would become close to-the theoretical limit. * “ an # neat Guna On re rr re iene ene ane de ere ne LT = es or Buds 256 DIRECT CURRENT CABLES Since there is no a.c. excitation, other than from harmonics, the dielectric losses will be much smaller with d.c. On the other hand, with a.c., stress distribution is determined by permittivity, which varies little with temperature, whereas with d.c. it is determined by conductivity, which varies greatly. This is an advantage in that the parts which are hotter, and thus the more conductive, will be relieved of stress. In cables, however, the core will be hotter than the sheath, so that the highest stress may be near the sheath, where slight defects are not so readily avoided (this problem of stress inversion is considered further in Section 14.4.2). 7 There is also with d.c. a very much greater chance of electro-chemical deterioration, due to the flow of unidirectional leakage current. The effect in this case is not to reduce the working or breakdown voltage of a new cable, but to tend to reduce its life; on the other hand, experience with d.c. capacitors has shown that dielectrics can be operated for adequate life-times with high, constant, unidirectional stresses. 14.3. Practical dielectrics Only two basic types of dielectric merit serious consideration for high voltage d.c. cables; these are impregnated-paper and polythene. Other dielectrics, such as vulcanised rubber and P.V.C., are available but are not considered to be sufficiently developed or having suitable characteristics for highly stressed and economic high voltage d.c. cables. 14.3.1 Impregnated paper This consists basically of insulating paper, dried and thoroughly impregnated with a suitable compound to give a high clectric strength. This type of dielectric is divisible into three further classes: (i) Mass and Pre-impregnated solid-type dielectric In the mass impregnated dielectric the conductor is insulated with unimpregnated paper and screened overall; it is subsequenily dried in vacuo and then impregnated with an insulating oil which is viscous but fluid down to room temperature. The cable is provided with a lead sheath to prevent water affecting the dielectric, and armouris provided if required. In the pre-impregnated type of cable the papers comprising the dielectric are dried and pre-impregnated in sheet form on a special machine with a compound which is solid throughout the working temperature range. Subsequently, the paper rolls are slit into spools and applied to the cable conductor under a molten compound having a high solidify- ing point. Both types of dielectric in this category are suitable only for voltages up to 33 kV a.c., with limited diclectrie stresses and temperature ratings. (ii) Gas-filled pre-impregnated dielectric This type of dielectric consists of pre-impregnated papers applied to the conductor in air with gaps between edges of adjacent turns of paper, screened over the dielectric and then sheathed and armoured as required. After the cable and joint terminations have been completely installed the dielectric is charged with nitrogen gas to a pressure of 200 Ib/sq in. the lead sheath being reinforced with thin metal tapes to withstand this pressure. The basis-of the design is that sufficient gas space is provided within the diclectric to cushion the compound expansion with respect to temperature, the nitrogen pressure then being applied with the diclectric to raise the ionisation stress rating of the gas-filled spaces [ eee “chap hees ecient oo sp ireetiemien ena ati Sei est ete PRACTICAL DIELECTRICS 257 Ibe to a high value. The pressure rise which occurs with increase of temperature in service is by within the capabilities of the reinforced sheath. ity This type of dielectric is used for alternating voltages from 33 kV up to about 275 kV; iter, the dielectric stress rating is 100 kV/cm, and the temperature rating is 85° C. The cable fl dielectric is non-draining up to the maximum service temperature and because of the diclec- f tric design no sheath distension occurs. rin (iii) Oil-filled impregnated paper In this case the conductor is lapped with unimpregnated papers and subsequently, after vacuum drying the whole cable, it is sheathed prior to finishing the drying in vacuo and impregnated with a mobile oil via the hollow conductor, or the padding spaces of a three- core cable. The cables are provided with oil conservators at the ends for transport to site. To allow for oil expansion with temperature changes and after the joints and sealing ends have been installed completely, oil reservoirs or conservators are connected at suitable points, such as the terminations. When the cable is heated the expanded oil flows longitu- dinally along the cable structure via.the hollow conductor, or via the padding spaces in a multi-core cable, to these reservoirs. The dielectric stress rating is of the order of 100 kV/ cm, since all the spaces between papers in the dielectric are filled with oil and theionisation stress is high; the temperature rating is of the order of 85° C, which is the physical limitation of the paper and oil. As in the case of the gas-filled cable the physical behaviour is per- fectly stable and no sheath distension occurs. ving, 2 14.3.2. Polythene™8 This dielectric, which is extruded, was initially used for high frequency communication cables on account of its excellent electrical properties and is now being rapidly developed for a.c. power cables; such cables have been used for a.c. voltages up to 11 kV with thick- nesses of dielectric similar to those used on impregnated paper cables. One advantage is that a lead sheath is not necessary since polythene is non-hygroscopic, although it is per- meable to water vapour and gases. The limitation to the stress rating resides in the fact that polythene is much more ionisa- tion-sensitive than paper; ionic discharges in any voids result in the melting of the dielectric boundaries and progressive deterioration then occurs. The temperature rating is limited to 60° C, at which temperature the polythene softens. Further improved grades of the material are being rapidly developed and increases in the temperature rating can be expected. From an economic standpoint, the insulation costs of polythene are probably twice that of oil impregnated paper, but saving may be introduced by the elimination of a watertight sheath. On this basis, the position is attractive for large single-core polythene cables com- pared with the equivalent paper insulated ones; some time ago an estimate was made and is shown in Figure 14.2, the assumption having been made that the load is carried by two cables, each operating at rated voltage to earth, the one positive and the other negative. The operating stress of the polythene insulated cables has been assumed to be 300 kV/cm. The thermal resistivity of polythene is considerably lower than those of the conventional cable-making materials, so that current ratings are much the same as for paper cable. In the most unfavourable case, the current ratings of polythene cables may be as much as 15 per cent lower than those of equivalent paper insulated cables. a cp a sr 3 OTT 7 lees - 7 nani Raina Akai ian iain nn cin i hat Leach ntti i canna 258 DIRECT CURRENT CABLES az Ss 8 [ [ [ [ &, ee | System comprising two cables, one positive and one egative to ground at stated e S Ss So awalt - statute mile, arbitrary unils ase Cable cost per meg s —— 250 600 750 1000 1250 Load in megawatts Figure 14.2 Economic study of polythene-insulated h.v.d.c. cables For transient overloads, the current ratings currently recommended are based on a maxi- mum temperature of 95° C and assume that the cable is carrying full load at the time of a fault. This leads to short-circuit ratings which are about 85 per cent of those of paper insulated types. Preliminary tests with cables having small conductors show that this rating is well within the capabilities of polythene cable. The impulse strength of polythene cable is somewhat higher than that of paper cable and mean values ranging from 1,650 to 2,500 kV per cm have been given. It has been observed that the falling off in strength with repeated pulses is somewhat more rapid than with paper cable and this arises from its sensitivity to ionisation. This may well be improved in the future. The standards of insulation resistance obtained with polythene cables are of quite a different order to those attained with other types of cable and it is important to stress that this insulation resistance will be maintained even though the polythene core is completely immersed in water and even when the cable is at the bottom of the sea and therefore subjected to severe water pressures. The permittivity of polythene (2.3 times that of air) is substan- tially lower than that of any other cable insulant and, therefore, offers some advantage over other forms of insulation. Polythene cables show up well in respect of mechanical properties. The modulus of polythene is low compared with that of metals so that its longitudinal expansion and I ne * nearer ExR | | { { | { { me DIELECTRIC STRESS CONSIDERATIONS 259 bending is controlled by the conductors. On the other hand its modulus is sufficiently high to avoid excessive strain under load. As befits a material so generally used as insula- tion for flexible high-frequency cables, polythene is robust and has a high elastic strain. Its abrasion resistance is adequate but not outstanding. Its ability to withstand heavy mech- anical loads is comparable to lead, and therefore it is usual to provide armouring to avoid accidental damage.48 An experimental 200 kV d.c. submarine cable is shown in Figure 14,3,080) 14.4 Dielectric stress considerations 14.4.1 Insulation resistance characteristics of dielectrics In d.c. cable dielectrics, the voltages and electrical stresses are distributed according to the insulation resistance values. Figure 14.4 shows the typical relation between these factors for an impregnated paper dielectric, whilst Figure 14.5 depicts the same factors for a polythene dielectric. As can be seen the insulation resistance of the impregnated paper is highly temperature- and stress-dependent whilst that of polythene is less so. In these Figures,* it should be emphasised that the values given are specific test values and considerable variations may exist for other samples. Rubber is somewhat less dependent generally, but P.V.C. is highly temperature-dependent. *Supplied by the courtesy of B.1.C. (Submarine) Cables Ltd. a | | Ta i Figure 14.3. Anexperimental 200 kV d.c. submarine cable with polythene insulation armouring for severe mechanical con- | | { : This cable has been made with double ) ditions i \. Li feel rd ap ret net mens wae d ted ol ue rec oeH il nda ai nit ae acne eles ts . 7 7 S . i Si thin ns ea A la hk Ph Bilal 260 DIRECT CURRENT CABLES Log resistivity Stress 204Vlem Stress /20kYfem Stress 250k em O 10 20 30 40 50 60 7O 80 $0 00 Temperature, Figure 14.4 Resistivity-temperature characteristics of an oil impregnated paper diclectric for different voltage stresses ot DIELECTRIC STRESS CONSIDERATIONS 261 14.4.2 Stress distribution and inversion of stress with temperature With no temperature gradient in the dielectric the insulation resistance values are uniform except for the limited variation with Stress; the electric stress value at any point is inversely proportional to the radius of the point in the diclectric, as for the a.c. case: av V & mae ee loge ROU sees cece eeeecnce (14.1) r dv : : ; where fe =Diclectric stress at radius x V=Voltage x=Radius of point considered in the dielectric r=Radius of conductor R=Outer radius of the dielectric The maximum stress is: Vv V war A ets 7 (14.2) r In general the voltage drop 8V across any elemental ring of dielectric, x in thicknes s, at radius x, is: k. 8x x where A=insulation resistance of the dielectric at the point (radius x) T,=the radial leakage current per unit length of cable k=a constant 6V= - AT, 465 awvik This gives the stress ai AT dete catek lll LLL LL (14.3) Considering temperature effects alone the conductor is at a higher temperature than the sheath when carrying load currents; this reduces the stress at the conductor and increases that at the screen because of the relative decrease of insulation resistance at the former com- pared with the latter. An increase of stress generally tends to reduce the insulation resis- tance value and hence to stabilise or limit the tendency for the stress to rise Or increase at any point. In order to determine the stress/radius relationship under different temperature conditions in a cable dielectric, the latter can be divided up into a number of concentric rings, of which the mean temperatures are determined from the known conductor current loadings, the ther- mal resistance values of the diclectiic and the medium in which the cable is laid, and the basic temperature values of the surroundings. The insulation resistance values corresponding to these temperatures are determined from Figure 14.4 and 14.5, at the average stress in the dielectric. These values are then divided by mean raditis concerned and the quotients Tepresent the stresses at the Corresponding Sections whence the Stress/radius relationship can be drawn. 262 DIRECT CURRENT CABLES Log resistivity SEo “ L Cl O 10 20 30 40 50 60 70 80 90 100 Temperature: °C See Figure 14.5 Resistivity-temperature charactetistics of a polythene dielectric for different voltage stresses ion } t i } DIELECTRIC STRESS CONSIDERATIONS 263 a LL e 45C ¥ w S S&S Ambient ; “2 0 O-/ 0-2 0-3 O4 pu. radial distance across the dielectric Figure 14.6 Distribution of voltage stress in polythene cables By repeating the calculations, on a trial and error basis, but using the insulation resistance valuc corresponding to the first determinations of the stress, further approximations can be made of the stress curve until a sufficiently accurate result is obtained. It is, of course, important to check that the average value of stress corresponds to that of the voltage and the geometry of the cable section. If an expression can be determined to relate the behaviour of the insulation resistance with temperature variation and stress then stress inversion may be treated analytically. Generally speaking, the determination of an expression to fit the test results is difficult. These mathematical processes may be carried out by the use of a computer, as mentioned in the Appendix to this Chapter. Figures 14.6 and 14.7 show typical voltage stress-radial distance characteristics for a single- core d.c. cable under different temperature gradients. The following important observations can be made from these figures: - ; 1 Gaia or 264 DIRECT CURRENT CABLES (i) An increase of temperature gradient increases appreciably the degree to which the stress is inverted. (ii) The higher the mean temperature of the range the greater the degree of inversion; although the effect is relatively small. 4 _R A : (iii) The greater the ratio - the less the inversion of stress. r Higher voltage cables have greater temperature gradients across the dielectric but have ; R . : greater ratios of z Generally speaking, the temperature effect predominates and the inversion of stress increases with the voltage rating of the cable. Clearly higher dielectric thermal resistivities also increase the effect. It is extremely difficult to measure accurately the stress distribution in a d.c. cable (see Appendix to this Chapter), and in order to determine the effects of this stress inversion phenomenon it is necessary to carry out stability tests at various ranges of temperature and voltages. : °. t 65C a, C Wh ¢ a Amb, & Imbrent Ss ‘S xs Q O-/ 0.2 0-3 O-4 p.u. radial distance across the dielectric Figure 14.7 Distribution of voltage stress in oil impregnated paper cables ~_ lectric enion . THERMAL CONSIDERATIONS AND LOSSES 265 14.4.3 Effect of gas pressure on dielectrics Where voids are present in the case of impregnated paper, gas filled and also polythene dielectrics, the effect of supcrimposed gas pressure is to raise the dielectric strength. The ionisation or discharge stress of a gas space varies as the pressure therein, and raising the gas pressure to, say 200 Ib/sq in. has a marked effect. In the case of gas-filled dielectric an increase of 25 per cent is obtained on the d.c. short-time break-down values. For polythene dielectrics, the application of pressure from the conductor to an external re- inforced dielectric tends to compress the voids as well as charging them with gas at a high pressure. Jn the case of oil-filled cables it is well known that the application of pressure increases the dielectric strength generally. Information on this subject is at present very incomplete. 14.5 Thermal considerations and losses For both impregnated paper and polythene dielectrics it can be shown that the thermal instability temperatures for economic values of d.c. stress for cables is of the order of 200° C and long before this temperature is reached the dielectrics would be damaged physically. So far as the physical properties of the dielectrics are concerned the thermal limitations are the same as those for d.c. cables. However, the effect of electrical stress inversion sets the limit to the temperature rating in all d.c. high-voltage cables. The losses which occur in d.c. cables are those of the conductor and the dielectric leak- age. The latter is small so far as its effect on cable rating and temperature generally are concerned. The ohmic conductor losses have no skin effects as in the a.c. case. Normally, on d.c. systems, short-circuits are limited to extremely short durations of time by automatic grid control. With the maximum conductor service temperature limitations imposed by Stress inversion, it can be assumed that no extra conductor area is necessitated by short- circuit considerations. 14.6 Cable design values 14.6.1 Temperature limitations The temperature limitations for various dielectrics so far as physical propertics are con- cerned can be assumed to be well outside those imposed by considerations of stress inver- sion. The temperature levels due to the latter will depend to a certain extent upon the stress value or diclectric thickness which is used, and in turn upon the economics of individual cases. . Another relevant factor is clearly the ratio of the overall radius of the diclectric compared with the conductor radius, and also the external thermal-resistance conditions. In the case of the latter factor, if the sheath operates as a relatively high temperature then clearly the conductor can operate at a correspondingly high temperature because the ratio of the insula- tion resistance of the dielectric at the temperature concerned is the governing factor. Generally speaking, for cables for 100 kV d.c. and upwards the maximum conductor temperatures can range from 45° to 65° C. 266 DIRECT CURRENT CABLES 14.6.2 Stress values Stability tests carried out at 1.5 to 2 times the normal working voltage, and with con- trolled heating cycles to limit the stress inversion, indicate that maximum stresses are of the order of 150 kV/cm (peak values) to 250 kV/cm (peak values). Gas-filled and oil-filled types have higher stress ratings than solid-type or polythene cables. 14.7 Economics of d.c. cables compared with a.c. cables The d.c. electric stress rating of the dielectric is approximately 2 to 2.5 times that of the a.c. rating. So far as current-rating is concerned, there are only the ohmic losses in the d.c. cable to be seriously considered compared with ohmic, dielectric and sheath in the a.c. case, but the temperature of the d.c. cable is limited by stress inversion to a value lower than that in the a.c. case. The net result is that a considerably greater power rating is obtained with d.c. cables compared with a.c. cables of the same size. It should be noted that d.c. is particularly applicable to long submarine cable trans- missions where the (cable/equipment) cost ratio is high. The cost of the d.c. cables is only a fraction of those for the a.c. cases and in addition much less physical space tends to be required in the waterway involved. Due consideration in the d.c. case must be given, however, to problems such as migration of impregnating compound, movement of the cables on the sea-bed due to too low a ratio of weight/diameter, etc:, under the effect of currents, which may affect the economic considerations. 14.8 Accessories for d.c. cables In the design of joints for impregnated paper d.c. cables the design parameters have been related to the equivalent a.c. rating with satisfactory results. “Fully screened” joints are recommended, in which the conductor is jointed by a smooth soldered ferrule of conventional design and where the insulation consists of hand applied paper tapes applied to match the cable insulation, which is profiled or stepped at each side of the joint. The maximum dielectric stress in the joint insulation is kept within about 75 per cent of that in the cable itself. The outer casings of joints may be similar to those of a.c. joints but it is possible to achieve considerable savings in the cost of joints for single core cables by using ferrous components which are not permissible with a.c. Outdoor sealing ends for d.c. cables are basically the same as for a.c. cables in that the cable end is housed within a compound filled porcelain container but certain design parameters have to be increased for satisfactory results. Fora 100 kV d.c. cable, for example, there would be a d.c. stability test voltage of about 200 kV and an impulse test voltage of about 350 kV. To withstand the impulse test it will be necessary to provide adequate stress shiclding with the sealing end. From the point of view of a stability test there are two important problems; firstly there is the suppression of all corona around the high voltage terminal which may be achieved by fitting an unusually large corona shield, and secondly, the electrostatic deposition of dust and other contamina- tion on the surface of the porcelain makes it esscutial to use porcelain insulators of adequate height and having an adequate protected length. Every case should be designed to its own special requirements. Polythene cables are a new development in d.c. termination, and for these there is no doubt that injection moulding techniques offer the best solution to the problem of jointing. Hang. fi 8 0 erent sl at aa PRACTICAL CABLE CONSTRUCTION 267 Some progress has also been made in the moulding of stress cones in sealing ends with regard to the porcclain design, etc., the remarks made above regarding impregnated paper cables apply equally well to polythene cables. 14.9 General Examples of cables already in use are shown in Figures 14.872 and 14.9.020) Two examples of cables built for experimental purposes on extra high-voltage d.c. systems‘8» are shown in Figures 14.10 (a) and (6). Since extensive investigations are being pursued ee ee } Figure 14.8 1COkV d.c. cables used in the Gotland scheme The conductor is of solid copper with a cross-sectional area of 90 mm? and the paper insulation is 7 mm thick. This thickness was chosen to provide at least 20 per cent margin over an impulse voltage of 425 kV. The paper is impregnated with thick oil: two lead sheaths are pro- vided and are protected with galvanised steel wire armouring. The outer diameter is 50 mm and the weight is 8.4 kg/mm length of normal cable: the corresponding data for the lengths with reinforced armouring, as shown in this figure, which are for cable terminations at the sea coast are 70mm and 15.6 kg/mm. At normal working voltage the field strength is no higher than 21 to 23 kV/mm. It is anticipated, however, that this is a con- servative design and this figure can be raised to slightly - over 30 kV/mm. alee? Los Lina lenius.f 268 DIRECT CURRENT CABLES Figure 14.9 Cross-section of 200 kV, 150A dic. underground cables used in Kashira-Moscow h.v.d.c. system The design shown is for underground use, and two such cables are employed on the system: the design for underwater use is similar except that the sheath is thicker and a continuous armouring of round wires has been used, Dimensions in mm. 1, aluminium conductor; 2, paper insulation 11.3 mm thick impregnated with oil-colophony compound; 3, metallised paper screen; 4, lead sheath; _5, polyvinylchloride tape; 6, rubber-cloth tape; 7, steel wire armouring; 8, layer of cable yarn. (at the time of going to press) into d.c. cables, it has been thought fit to append an expert’s summary of the situation in the form, by kind permission of the author, of M. Tellier’s | paper to the Berlin meeting of C.IL.G.R.E. Committee No. 10, 1959. i /2 | =~™ o~ ~ AA W NOGA ~NWAA HAYUDOO (2) ) Figure 14.10 Experimental cables for h.v.d.c. developed in the Soviet Union (a) -+400kV, 900A d.c. solid-type cable. 1, 1,000 mm2 aluminium conductor; 2, metallised unper- forated paper; 3, paper insulation; 4, lead sheath; 5, P.V.C. tapes; 6, bituminous compound; 7, pre- impregnated jute; 8, bituminous compound; 9, armour; 10, P.V.C. tape; 11, pre-impregnated jute; 12, bituminous compound. (5) -:400 kY, 900 A d.c. oil-filled pipe-type cable. 1, metallised unperforated paper; fa, perforated metallised paper and perforated copper tape; 2, 550 mm? copper conductor; 3, paper insulation; 4, oil, grade C-220; 5, D-shape skid wires; 6, stecl pipe; 7, protective coverings. il She no vosuch ivn for | oF is ! i} res round; el wire 7 — ww GS 8 Sait Appendix: Direct Current Cables BY R. TELLIER, ELecrriciveE DE FRANCE, PARIS (Translated by J. H. M. Sykes) Introduction The outlook for insulated cables in direct current systems has often been discussed in the technical literature. “s2) 483) The almost total absence of diclectric losses in insulating materials subjected to a d.c. voltage, as well as the low frequency of ionisation discharges, enables the designers of such cables to contemplate operation under gradients much steeper than those admissible for a.c. In addition, the phenomena associated with ageing of dielectric appear to be much less in evidence, and finally thermal instability should not introduce any special limitations, although according to tests carried out in France, such instability may occur under certain conditions. The new level of interest in d.c. transmission, evoked in many countries by technical progress in valve design, has set in motion a new range of practical studies relating to the behaviour of d.c. cables. The notes which follow are intended to set out some of the essen- tial features of the problems involved and the results already attained. Distribution of the voltage in the dielectric in a d.c. cable The application of a d.c. voltage to the cylindrical and supposedly homogeneous insula- tion of a cable, establishes a distribution of voltage gradient which is in principle, under steady state conditions, related to the resistivity of the insulating layers. This resistivity is particularly influenced by temperature and by the electric field.O8Y These influences have the result that in the general case the distribution of a d.c. voltage will be different from that prevailing under alternating or transient voltage, since in the last two cases the distribution is inherently determined by the capacitance of the insulating layers. This distribution is especially influenced by the thermal régime of the cable, and thus by the load transmitted.184) 85) G80) (187) (ss) A first approximation, which may be regarded as admissible in the study of the behaviour of cables with impregnated paper insulation, allows us to set out a law for the variation of resistivity of the insulating layers which takes the form e— aT TT) P=Po Po where p=the resistivity at T°C T=the temperature in °C of the point under consideration 1 oe : sete ‘ a= 7 =a characteristic of the insulation, in °C.-! ° Under these conditions a theoretical study clearly shows the conditions under which the load carried by the cable influences the distribution of potential, the determining factor being T re LIT A { foal Fuad ) re wo ao 270 DIRECT CURRENT CARLES the temperature drop across the insulant. At heavy loads, the maximal gradient may possibly appear on the external surface of the insulant, and may introduce a limitation of the power which the circuit can transmit,“8? particularly in the case of submarine cables where the cooling is efficient and where the maximum admissible temperature of the con- ductor may not be attained before the condition mentioned is found to exist. Actually, this law for the variation of the resistivity docs not take exact account of the phenomena which are brought into play, and a more complex law must be considered. An empirical relationship of the type: —(ke +aT) P=Po- € where € is the potential gradient at the point considered and k is a constant, has been proposed, but the solving of the problem Icads then to mathematical difficulties. A study of this problem with the aid of computing devices is at present in progress in France. In the case of polythene-insulated cables, the influence of the gradient on the resistivity appears to be too important to be neglected, even in a first approximation. This influence is such that it tends to make more uniform the distribution of the gradient in all circum- stances, and it also diminishes the change of stress due to the power transfer. An empirical law of variation of the resistivity of polythene of the form oT P= Po has been proposed,“87) and facilitated theoretical studies on the behaviour of cables insulated with this material. Indircet verification of the pheomenon of the transfer of the maximum gradient towards the external layers of the insulant when the temperature drop in the insulant of an impreg- nated paper cable is increasing, has been obtained by breakdown tests carried out under a relatively limited voltage and by creating in the insulant suitable temperature drops. How- ever, such tests must be carried out under carefully controlled conditions, so as to avoid thermal instability. The phenomenon may also be directly verified by arranging for electrodes to be placed in the dielectric, but the results obtained by certain workers in this ficld“8 appear to show that the distribution is falsified by the existence of space charges. Their divergent results may perhaps be explained by the presence of the clectrodes themselves.“190 The laws of the variation of the resistivity of insulants as proposed above are, of course, empirical in nature, and laws established ona physical basis and of a more general character may perhaps be established. The difference between the laws for the distribution of transient and d.c. voltage in the insulant of a cable tends to cause—apart from the above-mentioned influence of the power transmitted on the dielectric stress—the appearance of an additional stress in this insulant during rapid reversals of polarity. The following simple, but approximate, explanation may be offered. The gradient, for example in the neighbourhood of the conductor, under a voltage V, may be taken as having a value € , for a.c. and & ,,for d.c. The reversing of polarity on such a cable, previously subjected to a voltage of +V, corresponds to the superimposition of an additional voltage of —2V. During the transitory period, this latter voltage sets up ac a I uence i Gi {[ ilated i v pr lg a I iL- c avon if sltAv zsults o~ S = a TE nt L ower ulant fe E APPENDIX 271 an additional stress of —2 €, near to the conductor (assuming that the stress is pro- portional to the voltage) and the total stress becomes §& ; at this point, due to the stored charges in the insulant: : ej =—28,+8, or €;=2 ¢,—¢€, in absolute values. If ¢ °° &¢ (a paper cable carrying a load or a polycthylene cable), then: a This additional stress may cause a fault to develop in the cable if the coefficient of safety of the diclectric is not sufficient. Tests made in France have shown that in the case of polyethylene cables breakdowns may easily be caused in such conditions.490 Jmpregnated paper cables seem to be less sensitive to this phenomenon. Finally, mention should be made of the fact that when operating under a d.c. stress, the layers of paper—contrary to the a.c. condition—acquire a stress greater than that of the intervening layers of oil, and this factor is advantageous to the strength of the cable dielectric. Impreyuated paper cables D.c. cables manufactured up to the present time for industrial operation are all of the impregnated paper type, known as “‘solid” cables. The first d.c. cables were used not long after the beginning of the present century, in France (1907: Lyon-Moutiers) and in Great Britain (1910: Willesden-Ironbridge). The theoretical service gradients (220 kV/cm to 320 kV/cm) have been so far limited to low valiies in relation to the ascertained breakdown levels, which range from 1,000 kV/cm to 1,5¢0) kV/cm. The use of pressure cables has been envisaged, as scen in references “89 and 49), for certain projects. Tests carried out up to the present time in different countries (189) (187) (so) seem to indicate that in cables where there is a reduced coefficient of impregnation (for exampe, Solid type cable after several heating cycles or cables with pre-impregnated paper) there is @ reduction in the withstand level, in relation to the level attributable to a cable with nermal impregnation coeflicient. The use of a pressure of oil or gas enables the withstand level of these cables to be restored approxiately to the level of solid cables. However, the use of this pressure does not appear +o allow the increase in d.c. service voltage that it ensures with a.c. voltage, ‘where the ionisation level is thus sharply pushed back. A phenomenon is found similar to that of the withstaxd ability for impulse waves, which is only slightly affected by the pressure. It thes appears that the potentialities of pressure cables for d.c., in a comparison with a.c are somewhat less than had been thought in recent years. For this reason, the solid type cables, tue to their satisfactory performance and their simplicity, appear to be of special interesi for d.c. applications. ! Table 12 shows the main characteristics of various d.c. cables, made in different countries. The operation experience so far acquired with the existing cables has been favourable, but it Coes not yet permit of any certainty in fixing the admissible service gradient limits when ta<ing account of the load conditions and the possible over-voltages. ce oc oe 272 DIRECT CURRENT CABLES TABLE 12 Main characteristics of existing or projected d.c. cables France Cross- Sweden- U.S.S.R. Ntaly- New Lyon- Channel Gotland |__| Yugoslavia | Zealand Moutiers | Project 1954 Kashira 1958 1958 Installation 1907-1937 1959 (193) Moscow 1958 Project Project Project 1949 U9 (91) Type of cable .. Solid Solid Solid Solid Solid Liquid Solid Gas Voltage (KV). 100-150 100 100 220 400 400 250 Maximum current (A) Section of conduc- | 75 | 150 tor (mm?) +» | Cu | Cu 90 Cu 1,000 Al | S50 Cu 300 Thickness of the insulant(mm).. | 18 | 12 . 7 “ 18 16 Mean gradient (kV/cm)... | 80 | 130 250 Theoretical maxi- . mum gradient (kVjem) .... | 140 | 220 Calculated by formula Oil 150 200 900 900 800 In order to acquire further knowledge of the possibilitics of impregnated paper cables, additional investigations will be necessary, particularly on the following points: @ (b) (¢) (¢) (2) (f) (g) The mechanism of d.c. breakdown of taped dielectrics. Little information is presently available on this problem, and basic investigations are required for a better understanding of the phenomena involved.(200) The distribution of the stresses in the insulant, under both steady and transient conditions (thermal and electrical), in order to gain more knowledge of the practical influence of the modifications of the voltage distribution. The influence of the characteristics of the papers making up the cable and the nature of the impregnating material on the diclectric strength: studies already undertaken appear to show a certain advantage in favour of thin papers‘t89 and “fully impreg- nated” cables. The influence of the oil or gas pressure on the dielectric strength under direct voltage. The ageing of the diclectric of the cables under the effect of a steady stress over a long period (life curves); and the possible influence of the polarity and of the tem- perature. 7 The behaviour of cables during rapid discharges and in relations to overvoltages of different types. The problems relating to high voltage accessories, such as joints and sealing ends, which may justify designs different from the conventional a.c. designs. Spo teal dts = G 7 GC % ® cm om a TT Rig ries ean Sareea pr orange omen APPENDIX 273 It will be important that the value and the shape of service over-voltages on d.c. trans- mission projects should be well defined, in order to ascertain suitable test levels for such cables. Cables with thermoplastic insulation Among the synthetic insulating materials now available, only polyethylene has been actually used for experimental d.c. cables. Materials such as polyvinylchloride or synthetic elastomers have not appeared so far to be suitable for d.c. use, since they may give rise to electrolysis phenomena. However, according to recent tests carried out in Italy and Ger- many, special P.V.C. compounds could show good electrical properties under d.c. stress. New materials, such as polypropylene, have certain interesting features, but have not yet been subjected to sufficient tests to permit of their real possibilities being made clear. The attraction of polyethylene, particularly for the insulation of submarine cables, is undoubted, due to the simplicity and lightness of the resulting cables, which enables engi- neers to envisage cables for much greater depths than could be considered in connection with the more usual type of cables, and there is also a considerable degree of simplication in regard to possible repairs. However, while encouraging tests have been carried out in this field‘195) (198) (197) the relative techniques have not yet received the sanction of industrial experience. Polyethylenc-insulated cables, even when operating under limited gradients (150 kV/cm to 200 kV cm) appear to give rise, as with a.c., to partly unexplained breakdowns, which may be the result of local weaknesses which are difficult to detect. The use of these cables, which was envisaged for certain d.c. links, particularly for the Gotland scheme and for the cross-Channel cable, was not finally decided upon, in the light of the present state of the technique."9®) Polyethylene cables of various specifications, provided by various cable manufacturers, have been or are being subjected to tests in Sweden, where they are connected to the d.c. installation for the Gotland scheme. The results of these tests, when they become available, will constitute a most useful contribution to the study of cables of this kind. In regard to polyethylene cables generally, additional tests must still be undertaken in order to gain further knowledge of the characteristics of this type of insulation. The following points need particular investigation. (a) The mechanism of breakdown under d.c. stress, the study of which has also been so far very limited.“99) (6) The distribution of the stress in the insulation. (c) The ageing of the dielectric of the cables under the effect of a steady stress over a long period (life curves); and the possible influence of the polarity and temperature. (d) The ability to withstand polarity reversals and various types of over-vollages. (e) The problem relating to joints and sealing ends. { ei oe List of References ICatvertey, J. E.: “H.V.D.C. Power Trans- mission”, Lecture at Manchester College of Technology, 1953, Direct Current, 1953, Vol. 1, No. 7, p. 168. 2Errot, F. J. and Lorp Forrester: “Some Projects Favourable to D.C. Transmission and the Role of British Electrical Industry in relation thereto”, Proceedings LLE.E., 1949, 96, Part I, pp. 49-63. + 3Nacat, K.: “Operation Characteristics of A.C.-D.C.-A.C. Rectifier Invertor Set within A.C. Power System”, CIGRE Report No. 404, 1954. *ARMOLD, H. W. and Tuompson, J. R.: “‘Elec- tronic Frequency Conversion of Industrial Power”, Direct Current, 1953, Vol. 1, No. 6, pp. 154-156. ’ SBUSEMANN, F.: “D.C. Power Transmission Development by Siemens Schuckert Concern in Germany”, E.R.A. Report No. Z/T67, 1947. - ®Nexrasoy, A. M. and Pimenov, V. P.: “Deve- lopment Work in the U.S.S.R. on D.C. Power Transmission”, Direct Current, 1956, Vol. 2, No. 8, pp. 199-207. TEMRENSPERGER, Cu.: “High Voltage Direct J Current: The Swiss Viewpoint”, Direct Current, 1952, Vol. 1, No. 1, pp. 22-26, 8“The H.V.D.C. Transmission to Gotland”, Report No. 7313E of ASEA, Vasteras, Sweden. ®RATHSMAN, B. G. and Lamm, U.: “The Gotland H.V.D.C. Link: Present Progress”, Direct Current, 1952, Vol. 1, No. 1, pp. 2-6. loLamM, U.: “Progress in the Development of Power Transmission with H.V.D.C. in Sweden”, CIGRE Report No. 411, 1948. A1BUSEMANN, F.: “A. Model H.V.D.C. Trans- mission System”, E.R.A. Report No. V/T104, 1949. : 12Sayers, D. P., Lasorpe, M. E., and Lang, F. J.: “The Possibilitics of a Cross-Channel Power g Link Between the British and French Supply Systems”, Proceedings ILE.E., 1954, 101, Part I, pp. 284-308. W3Lane, F. J.: “The Cross-Channel Cable”, Direct Current, 1956, Vol. 3, No. 2, p. 39. J4BUSEMANN, F.: “Review of the Problems of H.V.D.C. Trensmission and Future Possibili- ties”, E.R.A. Report No. B/TI11, 1952. 15Ruscx, A.: “High Voltage Transmission Development in Sweden”, Proceedings LE.E., 1955, 102A, pp. 333-338. \8B6ckMAN, M.: “A Test Programme on Line Insulators for H.V.D.C.”, Direct Current, 1952, Vol. 1, No. 2, pp. 46-48. WRuscx, A., RATHSMAN, B. G., and GLIMSTEDT, U.: “The H.V.D.C. Power Transmission from Swedish Mainland to the Swedish Island of Gotland”, CIGRE Report No. 406, 1950. ISFALKINER-NUTTAL, G. R.: “Transmitting Small Amounts of Power over Relatively Long Distances: The Advantages of H.V.- D.C.”, Direct Current, 1952, Vol. 1, No. 1, pp. 6-10. J9FEINBERG, R.: “Frequency Changing with Mercury Arc Mutator”, J..E.E., 1939, 85 p. 531. 20Von Bertete, H.: “The Exploitation of Yugoslav Water Resources and the Pos- sibility of Using H.V.D.C. Transmission”, Direct Current, 1955, Vol. 2, No. 5, pp. 107-109, 21HoM, J. G.: “Direct Current Power Trans- mission”, Trans. A.LE.E., 1953, 72 Il, p. 1,114. *2BusEMANN, F.: “Artificial Commutation of Invertors”, E.R.A. Report No. B/TIO9, 1951. 23Lane, F. J., RATHSMAN, B. G., Law, U., and SmepsFELt, K. S.: “Comparison of. Trans- mission Costs for High Voltage A.C. and D.C. Systems”, CIGRE Report No. 417, 1956, aiso Direct Current, 1956, Vol. 3, No. 1, pp. 27-36. *4PALKINER-NUTTAL, G. R.: “Direct Current Overhead Line Transmission: Economic Advantages”, Direct Current, 1953, Vol. 1, No. 5, pp. 110-113. *>Borcguist, W.: “High Voltage D.C. Power Transmission”, C/IGRE Report No. 132, 1946. ®6BUSEMANN, F.: “Technical and Economical Comparison Between D.C. and A.C. Trans- mission of Power and Problems for Research and Development of D.C.”, E.R.A. Report No.:2Z/T77, 1949, also Water Power, Nov.- Dec., 1949, p. 225. 27ApAmson, C.: “A High Voltage Direct Current Laboratory at Manchester College of Science and Technology”, Direct Current, 1956, Vol. 3, No. 2, pp. 41-43. LIST OF REFERENCES 275 28BusEMANN, F.: “The Influence of the Number of Phases Used in a Converter on Har- monics”, Direct Current, 1954, Vol. 2, No. 3, pp. 65-72. 29SLEPIAN, J. and PAKALA, W. E.: “Backfires in Ignitrons in Series’, Trans. A.LE.E., 1941, 60, pp. 292-294. 30Von Bertece, H. and Tucker, R.: (“The v Design of High-Voltage High-Power Mercury- Are Converters”, Proceedings I.E.E., 1952, 99 I, pp. 555-581. 31UHLMANN, E.: “Alternating Voltage, Direct Volate Regulation and Power Factor of Converter Stations Operating on A.C. Systems of Finite Short-circuit Capacity”, Proceedings I.E.E., 1955, 102C, pp. 284-289. 32CaLverLeY, J. E.: “The Flow of Power and Reactive Components in Rectifier and Invertor Equipment”, English Electric Journal, March, 1954, p. 206, and June, 1954, p. 243. 33UHLMANN, E.: “The Operation of Several Phase Displaced Invertors on the Same Receiving Network”, Direct Current, 1953, Vol. 1, No. 5, pp. 106-110. 34Bacv, A. V., KRIcHENOVA, I. A., POLYAKOVv, V. E., Sincov, V. M., and Sropniku, V. Y.: “The Characteristics of Converting Plant”, Direct Current, 1952, Vol. 1, No. 5, pp. 130-131. 35LamM, U.: “Mercury Arc Converter Stations for H.V.D.C. Power Transmission”, CIGRE Report No, 133, 1946. 36READ, J. C.: “The Calculation of Rectifier and Invertor Performance Characteristics”, J. 1.E.E., 1945, 92 Il, pp. 495-509. 37SmirH, D. R.: “Characteristics of Six-phase Rectifiers from No-load to Short-circuit”, Engineer, Dec., 1955, p. 824. 38BUSEMANN, F.: “High Voltage D.C. Trans- mission, Part I, Economic Comparison Between D.C. and A.C. (Line Costs). Part II, Critical Review of Question of D.C. or A.C. Transmission (Line and Station Costs)”, E.R.A. Reports Nos. Z/T68, 1947, and Z/T69 1948, respectively. 39BUSEMANN, F.: “Starting of H.V.D.C. Trans- mission Lines with Converter Stations in Three-phase Bridge Connection”, E.R.A. Report No. B/TI01, 1951. 49BuseMANN, F.: “Operation of Short H.V.D.C. Overhead Lines”, E.R.A. Report No. B/T115, 1952. > 41Forsett, H.: “The Gotland D.C. Link: The Grid Control and Regulation Equipment”. Part I, Direct Current, 1955, Vol. 2, No. 5,+~ pp. 109-114. | Part II, Direct Current, 1955, Vol. 2, No. 7, pp. 166-170. 42Bora, L. F. and Encstrém, P. G.: “Applica- tion of the Transductor in Rectifier Tech- ~~ nique”, Direct Current, 1954, Vol. 2, No. 3, pp. 58-65. #®Cassiz, A. M.: “H.V.D.C. Power Trans- mission”, Lecture at College of Technology, Manchester, October, 1953, Direct Current, 1953, Vol. 1, No. 7, p. 166. 41BusEMANN, F.: “Hunting of Compounded Invertors for H.V.D.C. Transmission”, E.RA“ Report No. B/T 105, 1951. 45Miyamoro, S. and Naaal, K.: “The Stability and Frequency Variation of Three-phase Networks with Invertors and Synchronous Machines”, Direct Current, 1955, Vol. 2, No. 4, pp. 90-94. 46Cuam, E. J. and Derr, W. A.: “Automatic Control of Ignitron Rectifier Stations”, Trans. AJ.E.E., 1950, 69 11, 801-812. 47BUSEMANN, F.: “The Theory of the Control Problem of H.V.D.C. Transmission with Rectifiers and Invertors in Bridge Circuit”, E.R.A. Report No. Z/T74, 1948. 48BuUSEMANN, F.: “The Stability of Frequency of A.C. Systems Connected by D.C. Trans- mission Line’, E.R.A. Report No. Z/T75, 1948. 49BUSEMANN, F.: “Practical Reactive Power Requirements of H.V.D.C. Invertors with Natural Commutation”, Direct Current, 1952, Vol. 1, No. 2, pp. 35-38. 59°ScHULTZE, E.: “Possible Applications and Operating Characteristics of Power Trans- mission by H.V.D.C. in Comparison with A.C.”, Direct Current, 1956, Vol. 3, No. 2, pp. 52-57. S\BUSEMANN, F.: “Protection of H.V.D.C. Transmission Plants”, E.R.A. Report No. B/T108, 1951. 82SfAMAN, J. W. and Morton, L. W.: “A New Multiple High-speed Air Circuit-breaker for Mercury Are Rectifier Anode Circuits and Its Relation to the Arc-back Problem”, Trans. A.I.E.E., 1942, 61, pp. 788-796. 33GosLAND, L.: “Phenomenon Associated with H.V.D.C. Transmission”, Lecture at Man- $ chester College of Science and Technology, 1953, Direct Current, 1953, Vol. 1, No. 7, p. 166. 3{BuseMANN, F.: “H.V.D.C. Transmission from , Point of View of Planning Large Power ' System”, CIGRE Report No. 403, 1948. 55Kukekov, G. A. and Soroxin, P. G.: “Arc Quenching in Switchgear for Disconnecting Powerful D.C. Circuits at Very High Volt- ages”, Direct Current, 1954, Vol. 1, No. 8, pp. 213-216. Also Electrichestvo, 1953. No. 10, co nS lease og d 276 LIST OF REFERENCES So“A Particularly Effective Arrester for the Protection of D.C. Systems”, Brown Boveri Review, No. 4, April, 1951. 57BUSEMANN, F.: “Damping Circuits of Con- verters for H.V.D.C. Transmission”, E.R.A. Report No. B/T114, 1952. *SEpwarps, D. V. and Sait, E. K.: “Circuit Cushioning of Gas-filled, Grid-controlled Thyratrons”, Trans. A.LE.E., 1946, 65, pp. 640-643. >°Dittarb, J. K. and Batpwin, C. R.: (Jnr.): “Rectifier Arc-back Study on the Analogue Computer”, Trans. A.LE.E., 1954, 73 IL pp. 198-208. “Dortort, I. K.: “Diagnosis of Rectifier Ailments”, Trans. A.LE.E., 1949, 68 II, pp. 1,298-1,304, ®1BUSEMANN, F.: “Operation of an Invertor Supplying an A.C. System Without Syn- chronous Machines”, E.R.A. Report No. B/T112, 1952. . ®Witus, C. H.: “Harmonic Conimutation for Thyratron Invertors and Rectifiers”, Gen. Elec. Rev., 1932, 35, p. 632. Basar, G. J.and RABKIN, G.: “Grid Controlled Phase Advancer”, J. LLE.E., 1938, 82, p. 429. ®4TCHERVONENKISS, I. M.: “Ways to the Solution of Problems of H.V.D.C. Transmission”, Elektrochestvo, 1945, No. 3, and 1947, No. 1, also E.R.A. Report No. Z/T71, 1947. ®°Von Bertece, H.: ‘Forced Commutation for Power Factor Improvement in Converters”, Direct Current, 1952, Vol. 1, No. 1, pp. 14-21, ®6LamM, U.: ‘Forced Commutation in Mercury Are Converters”, Technical Achievements of ASEA Research, 1946. ®7BUSEMANN, F.: “Economic Supply of Reactive Power for H.V.D.C. Invertor Station”, Direct Current, 1954, Vol. 2, No. 1, pp. 8-15, also E.R.A. Report No. B/T1I18, 1954. ®SU.S. Patent No. 1,948,360. °9LUNDITOLM, R., SépeRvauy, C. E., Becktus, 1, and BéckmMaANn, M.: “D.C. Transmission with Return Current Through Earth”, CIGRE Report No. 401, 1948. *LuNpitoLM, R.: “Direct Current Through £ Earth”, CIGRE Report No. 134, 1946, 7 the Plant”, “1LUNDHOLM, R. and GOTHENBURG: “Return Current Through the Earth for H.V.D.C. Transmission”, Direct Current, 1953, Vol. 1, No. 4, pp. 79-86. “1 1pDen, L, Sven, S., and UHLMANN, E.: “The Gotland D.C. Link: The Layout of Part I, Direct Current, 1954, Vol. 2, No. 1, pp. 2-7. Part II, Direct Current, 1954, Vol. 2, No. 2, pp. 34-39. Suter, D. G.: “Direct Current Track Circuits 9 for Railways”, Direct Current, 1953, Vol. 1, No. 5, pp. 118-121. v 74BUSEMANN, F.: “The Magnetic Compass Error caused by D.C. Single-core. Sea Cables”, 2 E.R.A. Report No. B/T116, 1953, also Direct - Current, 1954, Vol. 1, No. 8, pp. 208-213. 75Wuttenean, S. and Raptey, W. G.: “Genera- tion and Flow of Harmonics in Transmission Systems”, Proceedings LE.E., 1949, 96 II, pp. 29-48, 7Raptey, W. G.: “Interference Between Power and Communication Circuits: Summary of Recent Information, 1926 to 1929”, J.LE.E., 1931, Vol. 69, No. 417, pp. 1,117-1,148, also E.R.A. Report No, M/T7, 1930. 77Rapiey, W. G. and WHITEHEAD, S.: “Recent Investigation on Telephone Interference”, J. LE.E., 1934, Vol. 74, No. 447, pp. 201-239, also E.R.A. Report No. M/T21, 1934. 78“Summary (to 1925) of European and American Data on Interference between Power and Communication Circuits”, JIE.E., 1930, Vol. 68, No. 401, pp. 587-641, also E.R.A, Report No. M/T3, 1929. Krewe, H. R. J: “Production, Flow and Effects of Harmonics in A.C. Transmission Networks”, E.R.A. Report No. MI/T99, 1948, also CIGRE Report No. 317, 1948. 8°Krewe, H. R. J.: “Electrostatic Induction between Power Lines and Telephone Lines”, Proceedings LLE.E., 1951, 98 I, pp. 121-127, also E.R.A. Report No. M/TIO9, 1950. 81“Directives Concerning the Protection of Communication Lines for Disturbing Actions of Electrical Power Lines”, C.C.LF. Rome Edition, revised at Oslo, 1938. 82FRIEDLANDER, E.: “Interactions between Har- monics, Transformer Saturation and Opera- tion of Rectifiers”, CIGRE Report No. 302, 1950. 83Marx, E. and GéscueL, H.: “Corona Losses at High Direct Voltages”, Elektrotechnische Zeilschrift, 1933, 54, p. 1,112. S'PassERIEUX, P.: “Comparative Measurements of A.C. and D.C. Corona Losses”, Direct Current, 1953, Vol. 1, No. 5, pp. 114-115. SSHENNING, B.: “Corona Effects in H.V.D.C. Lines”, CIGRE Report No. 135, 1946, also Direct Current, 1952, Vol. 1, No. 3, pp. 54-56, 86SLemon, G. R.: “Radio Influence from High- Voltage Corona”, Trans. ALE.E., 1949, 68 I, pp. 198-205. . S?CueN, K. and Borne, E. W.: “Testing of D.C. Interrupters on A.C. Test Circuits”, Trans. ALE.E., 1955, 74 (1, pp. 1,073-1,080. 88Von Berteve, H.: “The Continental Develop- ment of Single-Anode Mercury Are Rectifier Valves of High Power”, Proceedings LE.E., 1950, 97 IT, pp. 663-689. & oo : LIST OF REFERENCES S°LANGMUIR, I.: “Positive Ion Currents in the Positive Column of Mercury Arc”, G.E. Review, 1923, 26, p. 731. SOSLEPIAN, J.: “Theory of Current Transfer at Cathode of the Arc’, Physical Review, 1929, 27, p. 407. %Ssitn, C. G.: “The Mercury Arc Cathode”, Physical Review, 1942, 62, p. 663, alsoPhysical Review, 1946, 69, p. 96. ®2Von Berteve, H.: ‘“Cathode-spot Behaviour and the Thermal Control of the Emission Zone cr in Mercury Arc Rectifiers”, Proceedings LE.E., 1954, 101 Il, pp. 493-511, also Direct Current, 1954, Vol. 2, No. 1, pp. 15-18. ®3ROTHSTEIN, J.: “Cathode Vapour and Anchor- ing of the Cathode Spot”, Physical Review, 1949, 75, p. 1,336. ®tVon Bertece, H.: ‘Pool Surface and Cathode Spot Temperature of Mercury Cathodes as a Problem in Heat Conduction”, British Journal of Applied Physics, 1952, 3, p. 127. °5GUNTHER-SCHULZE: “Mercury Rectifier and Their Physical Foundations”, Engineering Progress, 251, Aug., 1925, ®6FRoomE, K. D.: “Current Densities of Free Moving Cathode Spot on Mercury”, British Journal of Applied Physics, 1953, 4, p. 91. ®?HERNQUIST, K. G.: “Discharge Mechanism of Mercury Pool Arcs”, Journal of Applied Physics, 1956, Vol. 27, No. 10, p. 1,226. ®8Von Bertetr, H.: “Current Densities of Free Moving Cathode Spots on Mercury”, British Journal of Applied Physics, 1952, 3, p. 358. *Tucker, R.: “A Review of the Development of Methods of Anchoring the Cathode Spot in Mercury Arc Rectifiers”, Direct Current, 1954, Vol. 2, No. 2, pp. 40-46. 100The Direct Current Study Committee of CIGRE, Mecting of June, 1958”, Direct Current, 1958, Vol. 4, No. 2, pp. 48-54. 10LVoN BerteLe, H,: “Steady Vapour Mercury Are Rectifier Valves”, Direct Current, 1952, Vol. 1, No. 2, pp. 43-45. 102BUSEMANN, F.: “H.V.D.C. Transmission”, Electrical Times, 1947, Vol. 111, p. 36, 98, 169. 103Von Berteve, H.: “The Design of Valves for H.V.D.C. Transmission”, Lecture at Man- “ chester College of Science and Technology, Summary in Direct Current, 1953, Vol. 1, No. 7, p. 166. 104*Rectifier © Transformer Characteristics”, A.ILE.E. Committee Report, Trans. A.LE.E., 1949, 68 II, pp. 1,305-1,311. 105PimENov, V. P., Posse, A. V., Remper, A. M., Roxorsan, S. S., and Turetski, V. E.: “The D.C. Transmission from Stalingrad = 277 Hydro-electric Station to Donbass”, Direct Current, 1957, Vol. 3, No. 4, pp. 106-112, translated from Elektricheskie, No. 11, 1956. 106]. amM, U.: “Operation Experience on the Gotland Scheme”, Lecture on H.V.D.C. Transmission at King’s College, London, summary in Direct Current, 1954, Vol. 2, No. 3, p. 72. 10°7BUSEMANN, F.: “Hunting of. Rectifiers with Strong Compounding for H.V.D.C. Trans-_L mission”, E.R.A. Report No. B/T104, 1951. 20SReIDER, A. M.: “Analysis of the Stability of the Control Systeras of the Kashira-Moscow Direct Current Transmission Line”, Direct ~ Current, 1957, Vol. 3, No. 7, pp. 227-240. 109*Proceedings of the Direct Current Study Com- mittee No. 10 of CIGRE”, Meeting at Lenin- grad, 1957. Part I, Direct Current, 1957, Vol. 3, No. 6, pp. 179-185. Part II, Direct Current, 1957, Vol. 3, No. 7, pp. 241-244. Part II, Direct Current, 1958, Vol. 3, No. 8, pp. 267-272. 110MeLGunov, N. M.: “High Voltage D.C. Power Transmission in the Soviet Union”, Direct Current, 1957, Vol. 3, No. 6, pp. 172-179. 11PimeNov, V. P.: “The Work of the Direct Current Institute”, Direct Current, 1957, Vol. 3, No. 6, pp. 185-191, M2HINGoRANI, N. G.: “Possible H.V.D.C. System Problems involving Switching”, Direct Current, 1958, Vol. 3, No. 8, pp. 245-247. 113K yuHe!, Hosutno: “Examples of Sea Return Circuit for Railways in Japan”, Railway Technical Research Institute, Japanese National Railways. Paper submitted to CIGRE Study Committee No. 10, 1958. M4BaceNov, S. A., Nocotsxy, K. K., AND Micuaitov, M. I.: “Corrosion of Under- ground Metal Constructions in Earth-Return D.C. Transmission”, Electricheskie Stantsii, 1956, No. 1. English translation: CIGRE Study Committce No. 10. 115PeTTERSSON, G. A., AHLGREN, L., and Forse, H.: “Telephone Interference and other Effects caused by the Gotland H.V.D.C. Transmission”, CIGRE Report No. 324, 1958. W6ViviAN, A. C. and Gerrarp, J. S.: “Cathodic Protection: Some basic Considerations”, Sf Direct Current, 1955, Vol. 2, No. 6, p. 138. U7De Brouwer, R.: “Cathodic Protection of Buried Metallic Structures”, CIGRE Report No. 205, 1948. UsSBrown, H. D. and Ssurn, J. J.: “Current and Voltage Wave Shape of Mercury Arc Rectifiers”, Trans. A.I.E.E., 1933, 52, p. 973. { foes wD lal 278 LIST OF REFERENCES 119Kosrenko, M. V., Mikuaitov, M. J., and CHERNYAEY, I. V.: “Interference from Three- phase Transmission Lines on Telecommunica- tion Circuits”, CIGRE Report No. 317, 1958. 29Scumipr, A. (Jr): “Capacitors in Power Systems with Rectifier Loads”, Trans. A.LE.E., 1953, 72 I, pp. 14-17. Also Direct Current, 1959, Vol. 4, No. 4, pp. 116-119. 121BoRNITZ, E., HOFFMANN, M., and Lerner, G.: “Harmonics in Electrical Systems arid their Reduction through Filter Circuits”, C/IGRE Report No. 304, 1958. i 122Pyntsov, A. M.: “The Calculation of Harmonics of Audio-Frequency Currents in D.C, Power Lines”, Direct Current 1958, Vol. 4, No.1, pp. 8-13. 1234. .E.E. Committee Reports: (1) “Inductive Co-ordination Aspects of Rectifier Installa- tions”, Trans. A.ILE.E., 1946, 65, pp. 417- 435. (2) “Inductive Co-ordination ASpects of Direct Current Systems supplied by Recti- fiers”, Trans. AJ.E.E., 1951, 701, pp. 1,034- 1,054. W2tMfarti, O. K. and Taytor, T. A.: ‘“‘Wave Shape of Thirty- and Sixty-phase Rectifier Groups”, Trans. A.LE.E., 1940, 59, pp. 218- 226. 225Buraey, F. 1, Gross, E. S., Levirsxt, E. K., Mankin, E. A., Posse, A. V., SAKOVITCH, A. A., and Touretsxt, V. E.: ‘Design Features of the Stalingrad-Donbass, 800 kV D.C. Line”, Report of CIGRE Study Com- mittee No. 10, 1958, also Direct Current, 1958, Vol. 4, No. 2, pp. 59-66. 126The Experimental-Industrial, 200 kV D.C. Transmission System, Kashira-Moscow”, CIGRE Report, 1956. 127HuLL, A. W. and Brown, H. D.: “Fundamental Electrical Properties of Mercury Vapour and Monoatomic Gases”, Trans. A.LE.E., 1934, 53, p. 1,435. 1228Hoyaux, M.: “A New Contribution to the Theory of the Mercury Are Rectifiers, Part I”, Direct Current, 1953, Vol. 1, No. 7, p. 192-200. 129Comprton, K. T. and Lamar, E. S.: ‘Potential Drop and Ionisation at the Mercury Arc Cathode”, Physical Review, 1931, 37, p. 1,069. Erratum in Physical Review, 1935, 48, p. 777. 130Sr. Jon, R. M. and Winans, J. G.: “Motion and Spectrum of Arc Cathode Spot in a Magnetic Field”, Physical Review, 1955, 98, pp. 1,664-1,671. 131Compron, K. T.: “On Theory of the Mercury Are”, Physical Review, 1931, 37, p. 1,077. 132TANBERG, R.: “Motion of an Electric Are in a Magnetic Ficld under Low Gas Pres- sures”, Letter to Nature, 1929, 124, p. 371. 133SmirH, C. G.: “Cathode Dark Space and Negative Glow of a Mercury Arc”, Physical Review, 1946, 69, p. 96. 134eRNQUIST, K. G.: “Emission Mechanism of Cold Arcs”, Physical Review, 1958, 102, p. 636, 135Rosson, A. E.: “The Electron Emission of the Cathode of a Cold Arc”, E.R.A. Report No. L/T331, 1955. “The Concentration of Currents at Are Cathodes”, E.R.A. Report No. L/T330, 1955. “The Cathode Spot: Some New Light on an Old Problem”, Direct Current, 1958, Vol. 4, No. 2, pp. 36-40. 136FEINBERG, R.: “Research in High Voltage Direct Current Transmission at Manchester College of Science and Technology”, Direct Current, 1958, Vol. 4, No. 3, pp. 81-83. 137STRATFOLD, P.: ‘The Mercury Arc Cathode Spot”, Thesis for the Degrce of M.Sc.Tech., October, 1958. Faculty of Technology, University * of Manchester (Manchester College of Science and Technology). 138CompToN, K. T. and LANGMurr, I.: “Electrical Discharges in Gases”, Review of Modern Physics, 1930, Vol. 2, Nos. 2 and 3, pp. 124, 129. 1397. aNGMuIR, I. and Motu-Smitu, H.: ‘Studies in Electric Discharges in Gases at Low Pressures”, General Electric Review, 1924, Vol. 27, pp. 449, 538, 616, 762, and 810. 140. AMM, U.: “The Gotland Scheme and H.V.D.C. Developments in Sweden”, Lecture at Manchester College of Science and Tech- nology, 1953, Direct Current, 1953, Vol. 1, No. 7, p. 177. WALE.E. Committee Report: “Mercury Are Power Converters in North America”, Trans. A.LE.E., 1948, 67 U, pp. 1,031-1,059. 1424...E.E. Committee Report: “Water Cooling Systems of Mercury Are Rectifiers,” Trans. ALE.E., 1953, 72 I, pp. 465-475. A43STemner, H. C., Zewner, J. L., and Zuvers, H. E.: “Pentode Ignitrons for Electric Power Converters”, Trans. A.LLE.E., 1944, 63, pp. 693-697, JHENGLISH ELecrric Co. Pustication: “Devel- opment in Mercury Converters for Large Industrial Drives”, Proceedings of Conference April, 1958, Stafford, England. M5SreptAN, J. and Lupwic, L. R.: “A new ing the Cathode of an Are”, Trans. A.LE.E., 1933, 52, pp. 693-698. TT o Are + Bisse ae, iviicat i of pas, v See sn eather LIST OF REFERENCES 279 446Housey, J. E. and Hucurs, G. N.: “Main- tenance of Rectifiers on Electrochemical Installations”, Trans. A.LE.E., 1946, 65, pp. 436-441. 47Hutt, A. W. and Brown, H. D.: “Mercury Arc Rectifier Research”, Trans. A.LE.E., 1931, SO, p. 744. 448ReAD, J. C.: “High Voltage Steel-tank Mer- cury Arc Rectifier Equipment for Radio Transmitters”, J. LEE., 1945, 92 II, pp. 453-468. 149STemNeER, H.C. and Price, H. N.: “A 400 amp Sealed Ignitron”, Trans. A.LE.E., 1946, 65, pp. 680-685. 150Kincron, K. H. and Lawton, E. J.: “The Relation of Residual Ionisation to Arc-back in Thyratrons”, General Electric Review, Nov., 1939, 42, p. 474. 151Von IssenvorrF, J., SCHENKEL, and SEELIGER: Entstchung und Bekampfung von Riick- zundiingen”, Wiss Siemens, 1930, 9, p. 73. 152Boyer, J. L. and Hacensick, C. G.: ‘“H.V. Ignitron Rectifiers and Invertors for Railroad Service”, Trans. A.LE.E., 1946, 65, pp. 463-470. 153The Cross-Channcl Cable: A Progress Report by a Special Correspondent”, Direct Current, 1958, Vol. 4, No. 2, pp. 41-47. 454Hewitt, P. C.: British Patent No. 10,672 of 1904 and No. 22,246 of 1913. 155WEINTRAUB, E.: “The Mercury Arc”, Trans. of American Electrochemical Society, 1905, 7, p. 277. 156AMILLAC, A.: Bulletin dela Société Alcascienne des constructions Mécaniques, 1924, 2, p. 63. 4187Von Issenvortr, J.: “Dic Verdampfung an der Kathode des Hg-Bogen”, Physikalische Zeit- Schrift, 1928, 29, p. 857. 158KopeL, E.: “Pressure of High Velocity Jets at Cathodes of a Mercury Arc”, Physical Review, 1930, 36, p. 1,636. 159Von Bertect, H.: “A new Phenomenon of Electron Emission from the Mercury Films. Letter to Nature, Oct, 1958, Vol. 182, No. 4,643, p. 1,148. 160Von BerteLe, H.: “The Cathode Spot in the Mercury Arc Rectifiers: A new approach to w the Theory of Control”, Direct Current, 1958, Vol. 4, No. 3, pp. 83-87. JIMULHERN, M. J.: “High Voltage Ignitron Rectifiers”, Trans. A.LE.E., 1950, 69 II, p. 913. 162 TeRsKIND, C. C. and Sremer, H. C.: “Rectifier Capacity”, Trans. A.LE.E., 1946, 65, pp. 667-669. 163 Herskinp, C. C. and Remscuetp, E. J.: Excita- tion, Control and Cooling of Ignitron Tubes”, Trans. A.I.E.E., 1946, 65, pp. 632-635. 454HeRSKIND, C. C. and Remscneip, E. J.: “Per- formance of Pumped Ignitron Rectifiers”, Trans. A.LE.E., 1948, 67 1, pp. 215-218. 165Marti, O. K.: “Excitron, Mercury Are Rectifiers”, Trans. ALE, 1940, 59, pp. 927-930. 166PaAKaLa, W. E. and Barren, W. B.: “Phase Occurrences of Arc-backs in High Current, Mercury Arc Rectifiers”, Trans. A.LE.E., 1940, 59, pp. 345-346. 167HuLt, A. W. and Exper, F. R.: “The Phase of Are-back”, Journal of Applied Physics, 1942, 13, pp. 171-178. 168WinoGRAD, H.: “Development of Excitron Type Rectifier”, Trans. AJE.E., 1944, 63, pp. 969-978. 169Boyer, J. L. and Coxaiaco, A. P.: “Sealed Ignitron Principle extended to larger Tubes”, Trans. A.IE.E., 1956, 75, 1, pp. 125-129. 170FreY, H. A.: “Insulator Surface Contamina- tion”, Trans. A.ILE.E., 1948, 67 Il, pp. 1,420- 1,425. 171Joun, W. J. and Sayers, F. M.: “Transmission Line Insulators under Deposit Conditions”, J. LEE., 1935, 77, pp. 629-662. 12Tayior, J. J.: “Insulators to withstand Air- borne Deposits”, Trans, A.LE.E., 1948, 67 II, pp. 1,436-1441. 173THompson, W. G.: “The Mechanism of the Contamination of Porcelain Insulators”, J. LEE., 1944, 91 I, pp. 317-327. 174GerTsik, A. K., Korsuntser, A. V., and Nikoxskn, N. K.: “The effect of Fouling on Insulators for H.V.D.C. Overhead Lines”, Direct Current, 1957, Vol. 3, No. 7, pp. 219- 226. 17%Witt, H.: “Insulation for H.V.D.C. Aerial Transmission Lines”, Report for CIGRE Study Committee No. 10, Meeting of Junc, 1958, Direct Current, 1958, Vol. 3, No. 2, pp. 48-54. 176YamapA, Y., Mira, N., and TAKEMuRA, T.: “Insulation for H.V.D.C. Acrial Transmission Lines”, Report for CIGRE Study Committee No. 10, Meeting of June, 1958. 177M ERKHALEY, S. D.: ‘Wet Flashover Character- istics of Insulators under D.C. Voltages”, Direct Current, 1959, Vol. 4, No. 4, pp. 99- 103. 278Forrest, J. S.: “The Electrical Propertics of Semi-conductor Ceramic Glaze”, Journal of Scientific Instruments, 1947, Vol. 24, No. 8. 178Recommendations for Mercury Arc Conver- ters”, Publication No. 84. First edition, 1957, of International Electrotechnical Com- mission, 1 Rue de Varembé, Geneva. 180MiLpNeR, R. C.: “Polythene-Insulated High Voltage D.C. Cables”, Direct Current, 1954, Vol. 1, No. 8, pp. 203-208. ae an tn nme kn en _— asa 280 LIST OF REFERENCES ISLWHITEHEAD, S.: Dielectric Breakdown of Solids, University Press, Oxford, 1951. 182DomMENACH, L.: “Cables for very high D.C. Voltages”, CIGRE Report No. 111, 1946. 183HANsON, B. and Buurstroy, B.: “Cables for the Transmission of Energy at High D.C. Voltages”, CIGRE Report No. 131, 1946. 418tSutron, C. T. W.: “Cables for High Voltage D.C.”, Direct Current, 1952, Vol. 1, No. 2, pp. 38-42. : 1S5Bracuine, S.: A High Voltage Cable for D.C.”, Electrichestvo, 1946, No. 2. 186PuGtiese, E.: “The Transverse Distribution of A.C, and D.C. Voltages in the Insulant of High Voltage Cables”, Bulletin del’ Association Suisse des Electriciens, 1950, Vol. 41, No. 25. 187TeLuER, R., CONSTANTIN, L., BRENAC, J. M., Oupw, J. M., and Bete, J.: “Research and Recent Progress in the Technique of D.C. and A.C. Cables in High Voltages”, CIGRE Report No, 212, 1958. 188STANNET, A. W. and Scnrorr, D. H.: “Tem- perature Coefficient of Resistivity of Poly- thylene and Oil Impregnated Paper”, Nature, 1957, Vol. 179. 189GoroneETZky, S.: “D.C. Cables for 200 kV to 400 kV”, CIGRE Report No. 206, 1958. WCIGRE, 1958: Discussion of Group 21, pp. 607-622. 29L“Submarine Link to be 250 kV D.C.”, Electrical World, 12th August, 1957. 192Gasser, O. and Hetp, Cu.: “Cable Problems with extra High Tension D.C. Power Trans- mission”, World Power Conference, 1954. 193ByurstRoM, B. and JOHANSSON, R.: “A 100 kV D.C. Cable”, CIGRE Report No. 112, 1954. 19tViseNTINI, M., Asta, A. and Trimant, F.: “Transmission of Electrical Energy by a D.C. Submarine Cable across the Adriatic”, CIGRE Report No. 210, 1958. 195RAZHENOV, S. A.: “The 220 kV Cable Installa- tion for the D.C. Transmission Scheme between Moscow and Kashira”, Direct Current, 1957, Vol. 3, No. 6, pp. 191-203. 196Painitscu, H.: “High Voltage Plastic Insulated Cables, Part IL”, CIGRE Report No. 205, 1956. 197Mitpner, R. C. and Humpriates, E. D.: Plastic Insulated Cables for D.C. and A.C.”, CIGRE Report No. 209, 1958. 1SHANSON, B.: “A Submarine Cable for 100 kV D.C. Power Transmission”, Trans, A.LE.E., 1954, 73 IIIA, pp. 599-605. 199K ircuin, D.-W. and Pratt, O. S.: “Treeingin Polythylene as a prelude to Breakdown”, Power Apparatus and Systems, June, 1958, p. 180. 200Mason, J. H.: Diclectric Breakdown in Solid Insulation. Progress in Dielectrics, Heywood and Co., London, 1959, 201Eppy, W. N. and Fenn, W. D.: “High Voltage D.C. Testing of Rubber Insulated Wire”, Trans. A.1.E.E., 1946, 65, pp. 676-678. 202VeiTH, H.: “The Dependence of the D.C. Resistance and Loss Angle of Paper on its Dryness and Temperature”, Frequenz 3, June and July, 1949, 2037 FiticH, K. and Ditcens, K.: “The Perfor- mance of Plastic (P.V.C.) Insulated Cables on D.C.”, Elektrizitats-wirtschaft, November, 1949, 204Howarp, P. R.: The effect of Electric Stress on the Life of Cables Incorporating a Polythene Dielectric”, Proceedings LE.E., 1951, 98 Il, pp. 365-370. 205BLaKELEY, P. W.: “Direct Current Power Transmission”, New Zealand Engineering, 15th Jan., 1958. 206TKepa, K.: ‘Technical Developments in Submarine Power Cables in Japan”, CIGRE Report No. 205, 1958. 207Perrenson, G. A. and JANCKE, G.: The Influence of High Voltage Power Lines on Telecom- munication and Low Voltage Installations due to Induction or other remote Effects, Published by the Royal Board of Swedish Tele-Com- munication and the Swedish State Power Board, 1957. 208MOLLERHOJ, J. S., Surron, C. T. W., and Morcan, A. M.: “Flat Pressure Cable for Submarine Installations”, CIGRE Report No. 201, 1958. 209PeeK, F. W.: “Law of Corona”, Part I, Trans. A.LE.E., 1911, pp. 1,889-1,965. Part I, Trans. AJE.E., 1912, pp. 1,051-1,092. Part III, Trans, A.LE.E., 1913, pp. 1,767-1785. 210WrnreHEAD, J. B. and Ler, F. W.: “The Electric Strength of Air under continuous Potentials and as influenced by Temperature”, Trans. AJLE.E., 1921, Vol. XL, pp. 1,201-1,308, 211FARWELL, S. P.: ‘The Corona Produced by * continuous Potentials”, Trans. A.LE.E., Vol. XXXiIf, 1914, pp. 1,631-1,671. *12TIKHODEEV, N. N.: “Calculation of the General Corona Initiating Voltages on D.C. Trans- mission Lines”, Elektrichestvo, Oct., 1957, No. 10, pp. 12-19. English translation No. M422, D.S.ILR. Lending Library, London. 213Voropev, A. V. and TikHopveev, N. N.: “Influence of Geometric Parameters on High Voltage D.C. Transmission Lincs on their Universal Sorona Characteristics”, Part I Unipolar Lines, Part IL Bipollar Lines. as LIST OF REFERENCES Zhurnal Tekhnicheskoi Fiziki, 1956, Vol. 26, No. 4, pp. 759-766 and 767-771. English translation from Science Museum Library, London. 24T;xHOpEEV, N. N.: “Effect of Geometric Parameters on the Generalised Corona Characteristics of High Voltage D.C. Lines”, Part III Lines with Horizontal Protective Cables, Zhurnal Tekhnicheskoi Fiziki, 1956, Vol. 26, No. 11, pp. 2,518-2,523. English translation from Science Museum Library London. 215MackENziE, D.: “Corona in Air at con- tinuous Potentials and Pressures lower than Atmosphere”, Physical Review, April, 1915, Vol. V. 26KUKEKOV, G.A.,SOROKIN,P.G.,andSHIPULINA, N. A.: “Switchgear for H.V.D.C. Lines’¢~ Direct Current, 1959, Vol. 4, No. 5, pp. 123- 126. 217Nexrasov, A. M. and Posse, A. V.: “Work Done in the Soviet Union on High-Voltage Long-Distance D.C. Power Transmission”, Ibid, pp. 515-521. 218K uxexov, G. A.: “A Very High Voltage D.C. Circuit-Breaker”. Papers of the Inter-College Conference on Long-Distance Transmission, Sect. III, Published by Leningrad Polytech. Inst., 1957, p. 77. 281 21sWittis, C. H., Beprorp, B. D., and Erper, F. R.: “Constant-Current D. C. Transmis- sion”. Trans. AJ.E.E. (Electrical Enginzer- ing), Vol. 54, January, 1935, pp. 102-108. 220BeprorD, B. D., E1per, F. R., and WiILLts, Cc. H.: “Power Transmission by Direct Current”. General Electric Review, Schenec- tady, N.Y., May, 1936, pp. 220-224. 221Posse, A. V.and Retork, A. M.: “On the Use of Series Connected Bridges and Tubes in the D.C. Transmission Scheme”. Izvestija Insti- tuta Postojanovo Toka, Leningrad, U.S.S.R., No. 3, 1958. 222NeYMAN, L. R., GLINTERNICK,S. R., EMELJANOV, A. V., and SHiputina, N. A.: “Group Con- nection of Tubes as a Method of Improving the Operating Reliability of Power Conver- ters”. Elektrichestvo, No. 6, 1956. 223R ewer, A. M.: “A Compounding Device for the Invertor in the Kashira-Moscow Trans- mission System”. Izvestija Instituta Posto- janovo Toka, No. 2, 1957. 224BUTAEY, F. L., Kuimov, N. S., Sakovitcu, A. A and STEPANOV, N. P.: “High-Voltage Conver- ter for D.C. Power Transmission”. Vestnik Elektropromyshiennosti, Moscow, U.S.S.R., No. 9, 1957. 225Breuer, G. D., Morack, M. M., Mortox, L. W., and Wooprow, C. A.: “D.C. Trans- mission: An American Viewpoint”. Trans. A.LE.E. (Power Apparatus and Systems), No. 43, August, 1959, pp. 504-515. Ao Pginaae SRST TTF am cir mm Index A Acration of valves, 240. Alternating current r.m.s. value, 26, 45-7. * Anchoring, 247-54, Angle of advance (£), 33-7. Angle of dcionisation, sce Deionisation angle. Angle of delay (a), 28-33. Angle of overlap (y), see Commutation angle. Arcbacks, 15, 20, 95, 97-9, 237, 243-7. Arc drop, 212, 220-1, 232. Arc through, sce Fire through. Are quenching, 20, 95, 99-100, 237, 242-3, 247. Artificial (or Forced) commutation, Chapter.8, Auxiliary supplics, 56-8, 101, 235, 238, 241. ‘ B Backfires, see Arcbacks. Batlle, 234, 237, 244, Bridge connection, merits of, 14-5. - working of Invertor, 31, 33-5. working of Rectifier, 12-4. Bridges in series, see Series connection of bridges. Bridges in parallel, sce Parallel connection of bridges. Bundling of conductors, 202-5. By-pass valve, 86-90, 104-6. Cc Cables, Chapter 14, 5, 7. Cathode fall space or zone, 218-9, 221. Cathode spot, Theories and characteristics, 221-5, 247-53. Cathode voltage drop, 221, 223-4, 25. Cathodic protection, 139, Characteristics of invertor, see Invertor charac- teristics. Characteristics of rectifier, see Rectifier charac- teristics. Circuit breakers, d.c., 9, 85, 101-6. Communication link, 72-3, 75, 83. Communication systems, interference with, 159- 67, 181-2. Commutation angle (y), 26-7, 30-5, 105-6. Commutation equivalent resistance, 28, 30, 32, 36, 37, 66-8. Commutation failure of invertor, 63, 92-6, 120. Commutation oscillations, 108-14. Commuutation reactance, 37-44, 243. Compass error, see Magnetic compass. Compensating reactors, 42, 44, 171. Compounding of rectifier and invertor, Chapter 5. Consecutive grid control, sce Constant extinction angle compounding. Constant current compounding or regulation, of rectifier, 62, 66-75, 88-9, 96-7, 101. of invertor, 73-5, 88-90, 97, 101, 103. Constant extinction angle compounding of inver- tor, 63-6, 93, 117-8. Control of d.c. system, Chapter 5. Converter circuits, types of, Chapter 2. Cooling of valves, 232-7. Corona, Chapter 12, 5, 192. Corona losses, 206-10. Corrosion, 133, 136, 138-9, 190. Critical grid voltage, 49, 226-7. Cross-Channel project, see English Channel project. Current regulation, see Constant current regula- tion, Current surges, 110, 112, 114, 237, 243. D Damping circuits, 69, 111-4. Degassing, 240, 242. Deionisation, theory and process, 225-6, 243-7. Deionisation time or angle (8,), 35, 63-6. Diametral connection, 11-2, 14. Dielectric, types of (for cables), 256-9. Double star connection, 12, 14. E Earth electrodes, 133-7. Earth return, Chapter 9, 4, 176, 182-3. Earthwire, 107, 209, 164. Economic distance for d.c. transmission, 9-10. Electrode sputtering, see Sputtering. Electro-magnetic induction, 160-7. Electro-osmosis, 134. Electro-static or Electric induction, 160-7. Emission theorics of cathode spot, see Cathode spot. English Channel project, 114, 141, 142, 176, 181, 215, 246, 247, 272. Equivalent circuit of, d.c. system, 67-9, 176-81. invertor, 36, 65-6. rectifier, 30, 32, 67. tation anode, 220, 237-40. Excitation failure, see Are quenching. wai | t hpetion f inver- hannel te 7 hee om INDEX 283 Field emission, 223-4. Fiim emission, 251-3. Filter circuits, 168-71, 181-2. Fire through, 20, 95, 100, 243-46. Forced commutation, see Artificial commutation. Frequency regulation, 79-80, 83. Fundamental current, 144-5, 149-56. G Generators, 8, 171. Glow discharge, 217-20. Gotland h.v.d.c. scheme, 6, 55, 64, 75, 76, 77, 83, 108, 113, 133, 135-6, 139, 169, 171, 182, 183, 247, 267, 273. Grading electrodes, see Intermediate electrodes. Grid bias, 49, 56. Grid blocking failure or Grid bias failure, see Fire through. Grid control circuits, Chapter 4, Grid, theory, 226-7. H Harmonics, Chapter 10. Harmonic interference, in the power systems, 256-9. in the communication circuits, 159-66, 181-3. Harmonics, method of reducing, 166-71, 181-3. Harmonics, on the a.c. side, 143-7]. on the d.c. side, 171-83. Hunting of d.c. system, 67-70. I Ignition failure, see Arc quenching. Ignition of valves, 220, 237-40. Ignitrons, 237-8, Insulation level, 107-8, 184-6, 196. Insulators, Chapter 11. Insulator types, Antifog, 186, 190. Post, 187. Stabilised, 199. Suspension, 185. Tension, 185. Interconnection of dic. lines, 9, 101-2, 106. different frequency systems, 7. Intermediate electrodes, 227-32. Inverse voltage, see Valve voltage. Invertor bridge, working of, sce Bridge connection. Invertor characteristics, 32-3, 36-7. Invertor compounding, see Constant current compounding and Constant deionisation angle compounding, K Kashira-Moscow Project, 120, 214, 268, 272. L Langmuir theory, 226. Lightning arrestors, 107-8. Line (d.c.) fault, 96-7, 101-6. Longitudinal induced harmonic voltage, 160-7. M Magnetic compass, interference with, 133, 139-42, Magnetic induction, 160-5. Margin of commutation, see Deionisation angle. Margin setting of current regulators, 74-5, 97. McLeod vacuum gauge, 241, 245. Mercury arc valves, Chapter 13. Misfire (failure to fire), see Arc quenching. Misfire (fire at wrong time), see Fire through. N Natural voltage characteristic of, Invertor, 66. Rectifier, 67. oO Oscillations, commutation, see Commutation oscillations in d.c. system. in the d.c. system, see Hunting of d.c. system. Output voltage, d.c., 24-30. Overvoltages, 106-14, 184-6. P Parallel connection of, bridges, 15-6, 18. valves, 20. Parallel operation of alternator and invertor, $2-3. Parameters of bridge rectifier and invertor, Chapter 3. Parasitic oscillations, see Commutation oscilla- tions. Paschen’s law, 219. Phase shifting circuits for grid control, 50-3, 60. Phase shifting of transformers, see Transformers. Photo-electric emission, 224. Pirani vacuum gauge, 241, 245. Polythene cables, 257-59, 262-3, 273. Potential dividers across, valves, 18, 20. electrodes, 232. Power factor, Chapter 7, Chapter 8, 28, 30, 47, 76-80, 154-5. -Power regulation, 72, 78-84. Protection, Chapter 6. a4 Bivssr 284 INDEX Pulse, types of, 49. Pulse forming circuits, 53-6. Pulse transformer, 58-9, 61. Pumping arrangement for a valve, 241. R Radio interference, 5, 110, 188, 198-9, 210. Railway signalling relays, interference with, 133, 137-8. . Reactive power requirement, Chapter 7, 8. Reactor, sce Smoothing reactor. Rectifier bridge, working of, sce Bridge connection. “Rectifier characteristics, 30-3. Regulation of d.c. system, Chapter 5, 8. Resonant commutation, 126-7. Reversing of d.c. power, 84. Ripple, see Harmonics. Russian h.v.d.c. schemes, see Kashira-Moscow project and Stalingrad-Donbass project. s Sea return, Chapter 9. Semiconductor rectifiers, 212. Serics connection of, Bridges, 16-8, 77, 88-90, 105, 110. Valves, 18, 20. . Skin effect, 5. Smoothing reactor, 67-9, 92, 97, 100, 104, 181-2, 179, Spot quenching, see Are quenching, Sputtering of clectrodes, 230, 241-2. Static capacitors, Chapter 8, 92, 119-20. Stability of d.c. lines, 4, 67-70. Stalingrad-Donbass h.v.d.c. project, 18, 107, 108, 114, 120, 186, 213. Swedish h.v.d.c. schemes, see Gotland project, and Trolihittan-Mellerud plant. eee Switchgear, d.c., see Circuit breakers. Synchronous condenser, 78-80, 92, 101, 119-20, 171. T Tanberg effect, 222, Telephone circuits, disturbance in, sce, Com- munication circuits. Telephone harmonic factor, 165. Telephone interference factor, 167. Thermionic emission, 223. Transformer, conncction, 15, 76-7, 143-5. phase shifting, 17, 166-8. rating, 14, 26. tappings or tap changing, 72-7. two secondary and one primary, 18-9, 43-5, winding ratio, 16. Transposition, 163-4, Transverse induced harmonic voltages, 160-5. Trolihattan-Mellerud h.v.d.c. experimental plant, 18, 52. Tuned circuits, sec Filter circuits. v Vacuum gauges, 241, 245. Vacuum pump, 241, 245. Valves, Chapter 13. Valve connections, Chapter 2. Valve voltage of, invertor, 37-42, 93. rectifier, 14-5, 45, 98-9, 108-10. Voltage divider circuits, 18-9, 232. Ww Weighting factor, 164-7. Naa nee nee are mee