ARCHIVO DE REFERENC GPrrro, BLED INDUS. INTERMACO SeCtoA O Femme O8, SucGESTED PRICE 50 CENTS ” THE WHYS OF THE THE BEHAVIOR OF H * TRANSFORMER wd 2 Y CONNECTIONS . GENERAL @ ELECTRICAn introduction "Te atten tgmsormat ere of course single phase, connected di- rectly across the two supply Hines. With the advent of the three-phase system, two alternative méthods of connection far transformers were offered—the delta fashion as the more obvious one based on tthe single-phase practice, and the Y fash~ {on as an intriguing new possibility, Some ‘body must have been thrilled to discover that the ¥ connection works, and that if fone wished to be different, he could even Connect one side of the transformer in Y. the other in delta. The ¥ connection must have appealed to the operators then, as it does now, because (a) it provides two different values of secondary voltage in- stead of one, and (b) it makes it possible to ground all three phases symmetrically ‘at a common point. Problems of the Y-¥ Connection It didn’t take the industry tong to discover one problem after another arising out of the Y-Y connection with the mary neutral floating. “The first problem discovered was that in three-phase banks of single-phase units fand in three-phase shell type units, though the line voltages conformed to turn ratio, the line-to-neutral_ voltages ‘were not $8% of the line voltages but about 68% at no load and diminished very rapidly when the bank was pro- ‘gressively loaded line to neutral ‘Oscillographie studies showed that with sine-wave voltages between line, the line to neutral voltages of these banks had ‘about 60% third harmonic component This explained the overvoltage at. no toad, because, at the effective values of voltages at different frequencies combine at right aneles, IEF X 5B) PART | to the behavior of transformer Y connections ‘The large voltage regulation of these tbanics for unbalanced line to neutral loads sas discovered to be due to the fact that, ‘whereas line-to-line loads cause a voltage Tegulation through the ordinary leakage impedanceof the bank, the line-to-neutral toads caused a voltage regulation through ‘their magnetizing reactance which jo gen- rally 100-1000 times as large as the Teakage reactance. ‘The three-phase unite with three-legged cotes were found to behave quite differ. tently in these matters. Their third har- monic voltages were negligible (according to the standards of those days); and the Fine to neutral voltages were practically 58%, and their voltage regulation for Toads to the neutral, though poor, was not altogether intolerable In present-day language, a balanced three-phate line-to-ine load causes volt lage regulation through the positive phase sequence leakage impedance X, of the bank; and @ line to neutral load through ‘2 combination of the positive and 2210 Zequence impedances of the bank, namely, (Xp 15%). In threesphase banks of Single phase units end in shelltype three- phase units, the Xj is the same as the Fraguetizing impedance, and therefore such banks were altogether unsuitable for heutral loading, However, in small three Tegaed core type units, X, was found to be of the order of 50%—100%6, and such unite could handle small high power factor line-to-neutral loads where the ‘quality of the service (constancy of velt- fage) was not very high grade. ‘Around 1910 the industry thought that it Understood the Y-Y connection well enough 40 a8 not to make a misapplication fof it. An operator hed a shelltype Y- connected autotransformer on his un- Grounded neutral system. He knew enough not to expect to load it to neutral, ‘but one day he wanted to ground the neutral end could see no harm in it. That 68% leg voltage (instead of the proper 84%) could not hurt the insulation of the fines or bushings in any way. So he closed the neutral to ground. The lines flashed land the breakers opened to the amaze- iment of the operator. The trouble was ‘raced to series resonance between the third-harmonic magnetizing reactance of the transformer and the line capacitance to ground, Although the third harmonic phenom- cena are more spectacular, yet Y connec: tion has also important 60-cycle problems land also impulse problems which would be well for an operator to understand. So, we will discuss in easy stages such ratters a8 these: In connection with the third harmonic, why third harmonic problems and not ‘other harmonica also; how to estimate the resonance condition; the effect of @ secondary (or tertiory) delta winding: reactance and size of a tertiary delta winding: the three-legged three-phase ‘core as the equivalent of a high reactance tertiary delta winding: residual third harmonic. In connection with 60-cyele problems, we may discuss the various utes of the hneutral; the requirements for satisfactory loading; the requirements for satisfactory ‘inversion phenomena in auto: In connection with transient phenom- ‘ena, impulse voltage behavior of wind- ings: differences between Y and delta; the problema of the autotransformer, and 0 forth.PART II Third-harmonic characteristics of single-phase units HY so much fuss over the third ‘harmonic in a transformer? Aren't there other harmonies in it too? The answer is, because we are interested in a three-phase system, In a fivesphase sys tem, the 5th harmonic would bother us; in an nephase system, the a’th. Te all comes about like this, Tn a symmetrical nephase star system with duplicate transformers (Fig. 1), ex ited sinuscidally and symmetrically, the ‘magnetizing currenis in the various phases will be alike in magaitude and distortion and displaced © (= 360/n) degrees apart. But in a symmetrical system, ifthe funda- mentals of two adjacent phases are © degrees apart, their third-harmonic com= ponents must be 3 @ degrees apart, the nth harmonic components n 0 degrees apart, etc. These two considerations—that the wave shapes must be alike and that the harmonies will be displaced proportional to their order—are found to be perfectly consistent with each other. Applying these principles to the various harmonics of the three-phase system of Fig. 2, we find the following results. (The even-numbered harmonics are left out because normally they do not exist in the steady-state mag- netizing currents, no matter how distorted these may be.) TABLE I PUASE ANGLES OF VARIOUS HARMONIC CONOR RC 0% HINGES HEE AVE. a ee ange gate te ta" tnd ih er ate nt eos raga ene Pte UBS Can vino op 80" te 2a ad Bees a nao Tanase pte Lay (Lazy Pont rw Some of the harmonics are seen to be threesphase, with a phase rotation or “sequence” like the fundamental (positive) and some negative; and some are single- phase, identical in all three phases, Thus, 4 tet, Th, 13th are 3-phasé with positive Sth, 11th, 17¢h are S-phase with negative rotation 3rd, 9th, 15th are i-phase with nor phate rotation Hence the adjective “zero-sequence™ for ‘the third harmonic and its multiples, ‘Those harmonic components of the three line currents which have the same frequency, and which are balanced and three-phate, flow freely in the lines, re- gardiess of the condition of the neutral Tead, because they add up to zero there. But those harmonic currents which are ine phase with each other in all three Tines, like the third-harmonic components, can- fot act as return to each other, and’ must neutral ie open as in Fig. 2, flow at all in the lines, ‘They are sup- pressed. The line currents of Fig. 2 there- they cannot fore cannot have any third-harmonic ‘The suppression of a current may be represented by the superposition on it of ‘an equal and opposite current, This super- posed current, acting like a magnetizing ‘current, will naturally produce and super~ pote on the circuit a voltage at its own, frequency. Thus it happens that the sup- pression of the third-harmonic current in ‘circuit like Fig. 2, consisting of single- phase units or @ three-phase’ shell-type ‘unit, produces large third-harmonic volt ages. ‘The magnitude of the suppressed third harmonic currents is about 2 half of that of the fundamental currents, and the resulting “third-harmonic voltages are about the same fraction (60% ~65%) of the fundamental voltage. riouRe 1, ‘with pica ao FIOURE 2. Theeesphate system With the help of the foregoing under. standing, we now can draw a oumber of inferences regarding the system of Fig. 2 1. ‘The thirdsharmonic, voltages induced in the primary coils A, B, C will be ine duced also, in the tertiary coils, a, by Gj and because the latier are in series, the voltage Eqn, across the open corner men, will be the sam of the three coil voltages. As the fundamental voltages cancel at m-n, Eqag will bea pure third and equal t0 3 x 65%, If we wanted pure third-harmonic voltage for some f supply for it, though fone of very high internal reactance and therefore of poor voltage regula- tion, 2. In Fig. 2 the third-harmonic voltages will appear in the line-to-neutral volt: ages AtoN, BtoN, and C to N, but rot in the line-torline voltages, At B, B10 C, and C to B, because in these later cases the third-harmonic voltages ‘of the two included phases oppose and cancel each other. 3, In general the third-harmonic voltage will appear alto from N to ground. Thit is based on the reasonable assumption that when we try to measure Eye by ‘connecting a voltmeter between N and G, there will be enough capacitance to ground from the lines A, Band C to complete the circuit of the voltmeter current. We may’ alto add that, whereas Eqn contains a fundamental and a third, Eyg will consist entirely of third-harmonic voltage, assuming that the system is perfectly symmetrical ‘There will be practically ino thicd+ harmonic in the Tine-to-ground vale- ages Exc, Eno Eeo. . If we should ground the neutral of the transformers (Pig. 3), leaving the neu teal of the source of supply isolated, Ene disappears completely, and Exoy Ege and Eg now exhibit the Full t harmonic voltages induced in the coils. ‘This can be dangerous, for each phase will be equivalent to a ciruit like that of Fig. 4, a third-harmonic voltage E; exciting & magnetizing impedance Xt in series with a capacitance reactance Xe. If Xi and Xe should be of the same order of value, there will be 2 eionance build-up ‘of voltage. The fact that X, it non-linear prevents ther HOURE 3, Tronsformer neutral grounded resonance from building up infinite voltages but not from building up troublesome ones, When the 60-cycle charging current of the line is of the order of one-third of the 60.cycle magnetizing current of the transformer, the values of X, and Xe will be alike for the third harmonic, ‘and this condition is not at all unusual, Large departures from resonance can still produce substantial harmonic in tensification. Let Xq be wice as large ‘as necessary for resonance: then the Tine-to-ground third-harmonic voltages (aerate Xe) will be twice Ey, and chus ‘equal to about 130% of the fundamen- ‘al. 6, Ifin Fig. 2 the'open corer m-n of the teary deta be cose the induced thd harmonic voltage Boon short Cieutedy and “appears fom. the primary ‘and the econdary aindgs Erwall except fora minute value due to the Teaage reactance betwoen the Windinge This howeeheuing. doe Mo produce any large eurent inthe eka but. jun the normal third farmoni exiing curent ose Viprimary dlaseandary transformers acre lot exhibit spore thirecharmonis treiing curent cictlsting in the delta qual fo about one hall of the exsing arent nthe primary. frihe dai winding had no other funeion or Guy thant circulate the thidehaconie cuten i hve raing FRB rere ars pra e ‘Cechaqen of the fal ood curren! of the Fespctoe nd eA eT PART Ill FIGURE 4, eh phowe would not be more than a few percent of that of the transformer, but as we shall see later, ordinarily itis subjected t0 other duties also, which necessitate a 30-50 per cent kva capacity in the tertiary, ‘The leakage reactance of a tertiary delta winding with respect to the other ‘windings can be very high and ic will sill reduce the residual third-harmonic voltage to a negligible value so far as quality of service and the safety of the windings and the connected apparatus are concerned; but telephone interference sets a more imit to residual harmonic voltages, demanding lower reactance tertiaries, and then single-phase shortcircuit currents demand greater current capacity in them. Singular properties of the core-type three-phase transformer unit a three-legged core, hi gular characteristics, both 60-cycle and third-harmonic, compared with a three- phase bank of single-phase units or the shell-type three-phase unit. At the bottom of these singular characteristics Ties the fact that the three-legged three phase unit, without a delta winding, behaves as if it had a concealed fictitious delta winding. Actually the tank acts a8 a loosely coupled equivalent delta winding ‘As a result of that, this type of a Y-¥ transformer exhibits 2% or less third harmonic voltage as against 60%-65% for the ¥-Y bank of single-phase units. There are scine important 60-cycle consequences also of this equivalent delta winding. The physical principle which this equivalent delta incorporates or explains is stated ‘more concisely in somewhat more sophis- ticated language by saying that the three- legged three-phese core construction reduces drastically the sero-phase- Sequence magnetizing impedance of a ¥-Y bank, Characteristics of the Zero-phase. ‘sequence Current As hinted before, @ current that is same magnitude and phase in each line— is called a zero-phase-sequence current (Fig. 1). Such @ current has to return through the neutral, making the neutral current three times that in each phase. As the normal balanced three-phase currents, whether positive or negative, do not flow in the neutral, we can identify @ ero ‘phase-sequence current in magnitude and —= 8 ee > Wau A Schamotie diaprem gf ‘sinding phase (and even wave shope) as one- third of the neutral current. Although the third-harmonic com- ponent of the exciting current is of this type, yet it may be clear from the defin- tion that 2 zero-phase-sequence quantity need not be limited to a particular frequency. In another installment of this article we shall deal with 60-cycle zero- phase sequence currents, which i another reason for belaboring the reader with this ‘matter whieh may sound academic but which actually developed out of practical engineering design work. Fig, 2 illustrates the zero-phase- sequence currents and fluxes of a bank of single-phase units. The solid arrows are currents, the dotted ones fuxes, The eur- rents are assumed to Row only in the ¥- connected primaries A. B, C; none in the secondary or tertiary windings (which therefore are not shown). The currents will be seen toact as magnetizing currents on the cores with the sac efectivencss as in normal operation.1. Although the fluxes in the three cores are in phase with each other, instead of the normal 120° angle between them, the three phases do not affect each other or modify the reactance offered by any one of the windings, Therefore, for the same current values, the int. edance offered to these currents is the normal magnetizing impedance of the transformer, 2s the magnetizing impedance of a ‘winding is the reciprocal of the mag, netizing current at unit (or 100%) voltage, for a transformer with .5% to 5% magnetizing current the corre. sponding impedance will be 20,000% to 200%, These values will of course be lower under saturation, For comparison with the foregoing, let Us consider now the three-phase core:type unit, on a three-legged core, illusteate Fig. 3. Here the dotted arcows (the fluxes) tay be seen to be lowing into a magnetic “blind alley." When they reach the yoke, they will find the magnetic highway to have ended at a wall, and they will have to jump off into the air to return to the bottom yoke to close their circuit. Where avin Fig. 2 the flux path was entirely in iron, ‘and the winding reactance was secordingly high, in Fig. 3 the introduc. tion of a big air gap into the path of the third-harmonic flux drastically reduces the reactance of the windings for these " FOURE 3. Current ond tux flow in three: Phase coretype unit, reelegued core 6 = ‘currents from the previously noted values (2000% to 20.00%) down to 509-2005. When the third harmonic current is suppressed, the resulting voltage is di. rectly proportional to the zero-phase. sequence impedance of the winding and hhence, in the 3-legged core designs, with ‘no conventional delta winding, the third harmonic voltage may be only # few per ‘cent compared with 60.68 pereent for a bank of single-phase unite having several thousand percent zero-phase-sequence impedance. If the foregoing has been of interest, the reader may wish to loole into these matters somewhat more quantitatively. We shall consider first the harmonic cur, rents circulating in both the conventional and the equivalent deltas; and then the resulting residual third-harmonic volt. ages, Harmonic Currents in the Deltas Suppose we have two banks of identical ¥-della windings, one set on single-phase cores, the other on a three-legged three. hase core. Assuming the normal leakage feactance of the windings as 10%, the bank of single-phase units will exhibit 10% zero-sequence reactance at 60 cycles, 30% at triple frequency; and the core. type three-phase unit will exhibit the Fesultant of the reactances of two delta windings in parallel, one conventional and. having 10% reactance, and the other equivalent and having (say) S0% react. ance, with a resultant vale of 8.3% at 60 cycles, 25% at triple frequency. The cur- rent in the conventional delta of the three-phase core-type unit will be 83% of. that in the delta of the single-phase units, the deficiency (17%) representing the iteulating current in the equivalent delta, of, what amounts to the same 'G the zero-sequence magnetizing ‘current of the core-type transformer, ‘The foregoing points to the conchision that, when an actual delta winding is pro. vided, there is not a great deal of differ: fence between the three-legged three. phase unit and the bank of single-phase units s0 far as the zero-sequence cireulat ing currents in their respective deltes are concerned, In the case of third-harmonic phenomena, this appronimate equality of circulating currents is in marked contrest to the big difference in the voltages in the absence of the deltas. Residual Third-harmonie Voltages. Let us see if we can calculate the fesidual voltages under various con. ditions, Fig, 5 shows one phase of bank and ‘tystallizes the problem. Let coil 2 be con- fected in delta and circulate the third ‘nie current. We wish to know the third-harmonic voltage Ey across coit 1 and connected in ¥. In this figure we find third-harmonic magnetomotive forces in three places; the magnetomotive force tops, Fain portion @ of the core, and Fy, in portion b of the core; and the magneto. motive force F, in coil 2 which is eirculat. ing the third harmonic mmf. Fy will ‘equal approximately the negative of the ‘counter magnetomotive forces, that is-— @,+FY). The total m4. drop along the closed path (a-+c) will then equal (Fy — Fy) which equals ~F,. The triple fre- quency flux in the leakage space ¢ will then be equal to F, times the permeance of region ¢, This latter is given directly by. the leakage reactance of the transformer, % AX» For the resulting voltage, this iy to be multiplied by 3 for triple frequency, 3X% IXuX% By Hee ror o ta Se i © FIOURE 4, Grounded neue Y-dlle winding ‘ > 1 i 1 cit] ; i 1 HOURE 5. Schematic view of one phere ks Coil twe la te be connected "et Geis Let us tabulate a few typical cases. We assume that the thied harmonic curtent is ‘one-half of the total, and that Fis ones thied of that of the total core ' We note that, ina ¥-delta transformer, with the leakage reactance up to 10% and the magnetizing current up to 5% the residual third-harmonic voltage it the ¥ is brought down below a quarter of one percent, ». Considering the bottom two items (50% and 200% reactance, represent ing’ ¥-Y connected 3-legged 3-phase units with equivalent high-reactance deltas), ‘the residual third-harmonic voltage in the ¥ is seen to be less than q“Lesleage Reactance ‘Third Harmonie Voltage in % of Fundamental Yito Delta % Xe When When When Mag. 11% | Mag. 1=295 Mag. 1=5% and Fy=0.17% | and F,=033% | end Fy=08% 5 By =0.0259% E,=0.05% By =0.1296 us 0.05% oie O24 ——-IOURE 6. Neuro grounded cos 5095 0.23% 05% 128 200% 1.00% 20% 48% 5% and generally tess than 2%, in contrast with the corresponding ¥-Y banks of single-phase units exhibiting 6095.68% third-harmonic voltage. Let us consider now two proposals: (a) to add an actual delta to the 3:legged ‘core-type unit; and (b) without the addi. tion of a delta, to ground the neutral of the ¥ on a grounded system, a. When an actual delta is added, this delta will in effect be in parallel with the squivalnt tale dels, and to, 8 ‘approximation the two deltas will ‘de the circulating third-harmonic current inversely as their reactances. Accordingly, in a Y-delta transformer fon @ 3-legged core having 10% normal reactance and an equivalent tank delts of 50% reactance, the real delta will carry $0/(50-+10) =83% of the total third, harmonic ampere-turns, and the equivalent tank delta the rest ‘Although in any case the circulating third-harmonie current are too small to be worth calculating so far as the safety fof the windings is concerned, they are important in case (b). b. If the neutral of a core-type three: phase Y-Y unit is grounded on a solidly grounded system (Fig. 6), the third-harmonic current has two parallel paths in which to flow: one is the equivalent delta, the other the ines and neutralof the system. Ifnow weassume ‘that the zero-phase-sequence imped- ‘ance of the supply system is 10% based fon the rated kva of the transformer, ¢ equivalent delta reactance as ne, for sng hat he rates PART IV 4 thee ow as 50% then $0/(504-10) 83% of the thirdshermonie current of the transformer will flow in the Tines and ground, In the ease of the bank of Y-Y Single phese units, the ground current would have been 100% of the third harmonic current of the transformer, instead of 83%, The reduction in the third-harmonic tground current accomplished by the Y-Y ccore-type unit is very small compared with the reduction in the third-harmonic voltage. In an ungrounded system, the neutral of a Y-¥ corentype three-phare unit ‘could be grounded with perfect impunity from thied-harmonic voltage troubles; but in a grounded system, about the same third-harmonic ground currents may be ‘expected from the three-phase core-type ‘unit as from a bank of single-phase unite (of any type), both banks in Y-¥ con: nection. Various sixty-cycle characteristics of Y transformer connections 'N THE preceding installments, four Y's were considered without formal listing as such, namely, isolated Y-Y, rounded Y-¥, isolated Y-delta, and ‘rounded Y-delta. There are more com- Plicated Y's also, like the 2ig-2ags, and Wwe shall have a few words about them, too. The 60-cyele characteristics of these connections to be considered here are the voltage strestes under line dis- turbances, the short-circuit currents under line faults, and the voltage regula tion under unbalanced loads to neutral. Y-¥ Connection, Isolated Neutral ‘The terminal characteristics of all transformer connections with isolated neutral are alike, and no external teste ccan distinguish between the Y and the delta or the other operative connections. ‘The inability of this connection to supply unbalanced fine to neutral loads has ‘already been pointed out in Part I. So this ‘case need be considered here no further. Y-Y Bank with Grounded Neviral 1, Supply System Isoleted Looking at Figure | it can be seen at 2 glance that, because the neutral is elec: tically tied down to ground, its poten- tial can never rise or sink, and therefore it would be possible to grade the insula. tion of the windings from zero (or nominal value) at N to full insulation test value at A, By C. This about exhausts possible favorable comments on this ease. As to the n of this system of operation, the following may be noted: (2). ‘The third harmonic hazards of this connection have already been pointed out in Part II, and need not be repeated here, ct YoY wlth supply syster oles 7(b)._ A little-known peculiarity of this connection is postible neutral inversion, ¢hat is, the neutral may go outside of the triangle of the line voltages, as illustrated in Figure 2. Figure 2a recognizes that Line 8 e ‘apetitence bonk, pore by the setup represents an inductance bank in parallel with a capacitance bank, phase by phase, Let all Xs be alike, ‘but one of the inductances (X',) be different from the other two (Xu, X1) for a reason soon to be explained. Let X, tbe smaller in ohme than Xo, and X'y, ‘greater than Xo. Then the net currents of phates A and C will be lagging, as the two X's draw more current than their mates X¢: while the net current of phase B will be leading. The normal posi tion of the neutral can not satisfy this condition, To see this clearly, consider the single-phase circuit of Figure 2b: 5 FIGURE 28, Single-phese deco the potential of point N' can not be be- tween those of A and B, but has to be cither above A or below B—the neutral N will be thrown out. Similarly, the neutral of Figure 2a will go outside of the triangle of the line voltages, as shown by the solid lines in Figure 2e, The dotted fines are the standard conditions for comparison. Now the current in line B ‘can return through lines A and © in orallel, This might be seen more easily with the help of the errows, The dotted arrows in Figure 2a are positive directions thought to be simpler for the present Purpose, and the solid arcows in Figure 2c are the vectors consistent with the assurmed positive direetions, ‘The reader ‘may choose his own arrowheads as he pleases, but those for Figure 2¢ must be ‘consistent with those for Figure 2a, T hope the reader feels that he is not interested in academic possibilities, that the three inductances should have been ‘assumed duplicates, because I feel that a way myself and, therefore, T assume the three transformers are duplicates with the further recognition that their reactances will vary with the voltage, Actually T hhad in mind traseformers that drew, at their normal voltage, a fraction of the charging current of the line; but under overexcitation, more than the chargi ccurcent of the line. Such system could stay in balance with the neutral in normal place, but is likely to be unstable; if any little transient condition should shift the neutral, it may keep going until it comes to balance as in Figure 2e and stick there ‘The requirement for this phenomenon, that at normal voltage the transformers should draw less magnetizing current ‘than the charging current of the lines, FIGURE 2C. Nevtrat outside voltage Wlonale Indicates that this trouble ia more likely to be experienced with grounded ¥-¥ potential transformer banks on moderate voltage short isolated systems. That thas been the experience, but the danger- ‘us proportions can also happen with distribution or even power transformers, Tf'these have not been experienced, it ‘must be because a grounded Y-Y power transformer bank on an isolated system ‘must be a very rare thing (©. On such a system, a Tine fault on fone phase, such as at B (Figure 3a), shorts the 60-cycle voltage of that phase, —B to N; and the voltage diagram of the windings changes from the solid lines of Figure 3a to the dotted lines, shown ‘more clearly in Figure 3b in solid lines, where we sce that the unfaulted legs take over the line voltages in open-delta fashion, representing 73% overexcitation. ‘The secondary (a, b, ¢) also is in open delta, with fullsline voltages across wind: ings that were formerly the lege of the Y. Such overexci ‘exciting current several times as large as the full-load current. Tt would be unfortunate indeed if the breakers were ‘on the secondary side (a, b, ¢), with only necting switches at” A, B, C, in which case the primary windings could ct roasted out, as this exciting current, even though very large, may not trip the breakers at the source. To" protect FIGURE A. Dotted linet shew caused by foull on phase 8 such a bank, it would be necessary to use differential relays arranged to short circuit the primary lines in ease of trouble, 30 as to force the supply station breakers to trip. But if we are reconciled to tri ping those breakers, the more appropriate transformer connection would be the grounded Y-delta or Y-Y delta because, addition to the benefit of graded insulation, we can also secure freedom from the 173% voltage stress we have noted, If, in Figure 3a, A-BC is a core-type three-phase unit, it will behave like a high-seactance Y-delta transformer and draw a larger fault current than a bank of single-phase units would do, While the latter draws only a (large) magnetizing current, the former would draw in addi tion a Small short-circuited current, (@), The neutral of the (a, Figure 1) may be eithe rounded, If it js isolated, no particular problems are presented other than those ‘of any isolated system, If the neutral is grounded, then a ground on the lines a, b, ¢ will have the same effect as @ ground on the lines A, B, C, shorting ‘out that phase of the bank’ and bringin about the same condition of overexcit. tion of the remaining phases as shown in Figure 3b. If the primary impressed volt- ‘ge is maintained, the secondary output voltage also will be maintained across the lines, even though the overexcited transformers may be roasti Af the secondary neutral it grounded, the purpose might be to opprate a four. wire system. Then an unbalanced circuit in the neutral will shift the neutral through the zero-sequence impedance drop of the bank. Such @ current can hhave any phase angle whatever, depend- {ing not oniy on the power factor of the losd, but also on the phase or combina- tion of phases from which itis being sup- plied. The worst situation {s when I, is A right angles to one of the phase volt. ages, the leading current being as ob- Jectionable as the lagging. Figure 4 shows 1, as lagging 90° behind Eq, The resultoy FLOURE 38. Soll tints show unfault staking over tin repent ing reactive voltage drop will be parallel to Eq, and will subtract from it arith: matically, n moving to n', s0 that the new phase voltages under’ such a load will Be Eyal, Byel; Ex! the change in Eyy being the worst, Let us estimate ite mag tude. AAs an upper figure for the neutral cur- rent, it is logical to assume that the neu- tral current entering the’ bank is equal to the rated line current of the bank, and FIoURE 4. then T, (=T,/3) will be 39%. As for the value of the zero-sequence impedance, in a bank of single-phase units (or shell-type Uhree:phase units), the minimum zero- ‘sequence impedance may be taken as 100% (ignoring saturation), in which case the neutral shift will be 333%, That is, the bank can not deliver such a load, 8 this exceeds the line-to-neutral short. circuit current of the bank, Even though the resulting overexcitation and satura tion will greatly increase the current, it may be clear enough that this is not a permissible operation, In the case of the coretype three: phase unit, we may assume Z, as 50%, fn which case the neutral shift will be 16.6%. If I, is lagging as shown in Figure 3, Ey will drop 16.6%; and if 1, is leading, Ey, will be boosted 16.696 Both cases ate intolerable. However, if the neutral current does not exceed 25% of the line current, the voltage regula: tion will be 4.1%, which then might be defended as tolerable as an occasional happening. T believe this has been tried more than once on 4 very small scale and gotten by in some cases, abandoned in others. 2, Both System and Y-¥ Transformers Grounded Considering Figure 5, we note that each phase of the transformer bank is directly associated with a corresponding hate of the source, and such operation should be foolproof. Well, it is-almost, ‘Such a system is free frog voltage rites due to neutral shift under line-to-ground. faults; and the secondary can handle a four-wire distribution system with nor ‘mal voltage regulation. However, both neutrals must be carried along on fourth wise both third-harmonic and 60-cycle currents will Row in the ground, with possible telephone interference and other troubles, The advantages of this connec- tion can be obtained and its tages avoided more or less by the and the Y-¥ delta connections, which we shall consider now, Y-Delta In a ¥-delta bank, whether the ¥ is primary or secondary, the grounding of the neutral of the ¥ permits the grading fof the transformer insulation and alto tends to stabilize the 60-cycle potentials ‘of ts terminals (Figure 6a), but its efficacy for the latter purpose depends on several ‘considerations ee FIGURE 5. System and trensforms When the source is effectively grounded, 4 line fault eliminates one phase voltage regardless of the connections of the step: down bank, and the voltage diagram becomes as’shown in Figure 6b: some- thing between A-BI-C ‘and AUBIC, depending on how closely the different Phases are coupled at the source. If the ¥ is primary, with its neutral Grounded, and the source is isolated, and if the souree impedance is negligible compared with the transformer imped- ance, then in cate of a linetto-ground HOURE 6A. va ta bon, ¥ nee fault, the primary voltages are main- ied and a voltage diagram like that of Figure 3b results, This points up the fact that an isolated aystem can not be effectively grounded by grounding the neutral of a small ¥-delta bank capable ‘of drawing only a small short-circuit current in cate of aline-to-ground fault. then the voltage diagram will be inter: mediate between AUBLC) and A.B.C ‘Figure 6c), the former representing the most effectively grounded system, the latter a poorly grounded (or isolated system). We may well asl, how dangerous is ABC (Figure 62)? How necessary is it to insist on approaching the AMBLC! perfection? ‘The sixty-cycle overvltages, amount- ing at the most to 173% of normal, are generally well within the insulation strength of the transformers, as all units receive a high-potential test of 3.46 X the leg voltage -F 1000 volts. The voltage stresses are not critical, but the following conditions at least must be satisfied The fault current must be eapable of burning off minor faults, and in case of their persistence, to trip the breakers at ‘the source, The kva rating of a grounding bank answering this requirement will be a substantial fraction of the source kva. Tt may also be seen that where the sixty-cycle overvoltages are dependably low, the lightning arresters can be se- ected for a correspondingly lower volt- age, thus affording better lightning pro- tection. In practice the arrester rating ‘can not be lowered down to S8% for two reasons. First, the short circuiting ‘of one phase to ground may increase the voltages across the other phases to ground. aeOf course this will not be anything like that in Y-¥ connection, but still it may bbe substantial ax dynamic voltage for the lightning arresters. Second, when breakers open a large load, the supply voltage may rise considerably untit volt. ‘age regulators can it back, and this also needs to be taken into account in the choice of lightning arresters, It appears that the appropriate lightning arrester levels for grounded systems vary, generally from 65% to 80% of the rated Une-tortine voltage. If the primary delte, and the secondary grounded Y, the ¥ secondary can furnish a four-wire distribution system with not much more than normal voltage regulation. Whether ‘or not the primary system and winding are grounded is immaterial; but economies ‘ate possible by grounding these. This leads us to the Y-delta-¥ connect Y-Delta-Y Banks The advantages of the Y-deta-¥ connection (Figure 7) are well-known: both Y's can be grounded and graded; ther ¥ can handle loads to neutral with, ‘normal regulation; the primary and secondary voltages are in phase with each other (in contrast to Yedelta); third ‘harmonic voltages are so emall that very refined measurements would be needed to detect them at all; the delta can be used for a tertiary load at ite own volt. ‘age; and the reactance of the delta can ¥ conntetion showing linestovground fall on primary Tina B PART V be adjusted to be different from th of the Y's, so that the bank will offer a desired reactance to line-to-ground faults, and another desired reactance to loads fand three-phase short circuits, Ie is customary to make the kva rating of the delta winding 35%-50% of the primary. This is dictated by the short. Gircuit duty imposed on the tertiary winding, Figure 7a illustrates @ line-to- ‘ground fault on the primary line B. The currents are drawn on the conservative assumption that the source is either isolated or so far away that all of the fault current in the ground flows into the transformer neutral. We see here that the short-circuit ampere turns in the delta are equal to those in the primary, and if the short-cireuit current must be large enough to trip the line breakers after the eet time delay, no 5 or 10 percent delta wit! do. Figure 7b illustrates a fault on the ‘secondary line b. The duty on the delta ow is one third of the maximum in the secondary, and one half of the maximum in the primary. Untess the delta is to ‘be the secondary or a major winding like it, an excellent compromise is effected by making the effective reactance of the bank for line-to-ground fault twice that of the major windings to three-phase shorts, and providing the delta with 50% capacity. This makes possible a rugged tertiary winding. The Y with the ground-fault neutralizer [8 THE catty days, isolated systems were more common than the grounded ‘ones on the theory that the isolated ays: tem would provide better continuity: of service, because a single line-tovgroand fault Would not trip the breakers ax Tt does in solidly grounded system. Ex: perience, however, showed that fre. uently ‘one line faule was followed by. another or a failure, and the breakers di trip after all. In the grounded system, the brealeers would have tripped without the second fault or the apparatus failure, 10 {and saved the failure. One explanation of this peculiar behavior of the isolated system is that a fine fault on such a system constitutes an unsteady arcing sound, with a capacitance discharge Current oscillating at the natural fre: quency of the circuit, and by restrikes could build up four’to five times the normal voltage on the lines.* Such per. re a oman, Gomice eel Ae Ee ett et ‘Asean colatoee sistent oscillations could obviously result in damage to the connected apparatus, This (and other considerations) led to wide acceptance of grounded-neutral operation. Meanwhile, invention took a ‘hand in this situation. Professor Petersen of Germany reasoned thet, if the arcing. ground current can be rendered non oscillating, this kind of a voltage build-up will be avoided, and then isolated system operation becomes practical. His solution (of the problem took the following simple form,‘Arrows rapreseat the symmetrical throo- Cg = phase charging currents of the lines (0 One Solution Figure 12 ifustrates the capacitance charging currents of a normal system. ‘The capacitance current of each line to ‘ground returns through the capacitances, of the other two lines, without requiring any current in the neutral, as for any ‘ymmetrical three-phase load. In Figure 1b, line Chas become grounded through a fault f, This removes the voltage across its capacitance C,. No current will now flow through C, for lack of voltage across it, but currents can flow through f, and we note that I, and Ty now return through this fault. These currents are in quadrature with the voltage producing them. Therefore, when the currents are zero, the voltages sre maximum, and any tendency to make the arc go out as the currents are passing ‘through zero brings on a large recovery voltage and possible restriking, giving ise to an oscillation between the series inductance and shunt capacitance of the lines. ‘The leading currents 14 and Tp could bbe neutralized at the fault by an induc tive current appropriately introduced into the circuit as shown in Figure te Here, Xp is an inductance connected be- tween neutral and ground, and its lagging current (Ip) returns through f. If it is of the right magnitude (that is, 1p= —Iy =Iy), it can neutralize 1 and Ip at f The arrow for Tp is shown opposite to those of Iq and Ig, because lagging and leading cutrents are naturally opposite: ‘The only current that can not be neutral- ized at f is the power component of the currents. But this component is generally very small; and, what is more important, when a power component of current is passing through zero, 0 is its voltage through zero, and the are ean go ithout bringing on a substantial re- covery voltage to cause a recurrent oscil- lation. This reactor (Xp) is appropriately, called a ground-fault neutralizer. ‘The maximum voltages to ground are practically limited to 1.73 times normal AAs the charging current and kva of the Tines are generally small, the current and. kya of the reactor to neutralize them are also correspondingly small, so that the cost of the equipment can not be pro- hibitive Neutralization requires that Xp be ads justed (tuned) to the capacitance of the system, and, therefore, a ground-fault neutralizer is built adjustable, ‘As this method of operation involves ‘connection between the neutral a1 ground, it becomes necessary to see to it that a thiedsharmonie voltage does not set impressed on the ground-fault neue tralizer, For such a voltage, Xp would be in series with the capacitances of the uunfaulted lines and would thus invite third-harmonic series,resonance troubles, So far as the 60-cycle voltages are con- cerned, Xp is, in effect, in parallel with the capacitances, and any resonance is parallel resonance, which is utilized to 0b the current from the arc, It is for this reason that Figure le shows a delta winding on the supply transformer bank. Selecting « System It is not the thought of this writer to advocate this particular system of opera- tion as ageinst grounded-neutral opera. tion. For a preference, an operator will aturally weigh many considerations, may be different in different cases. ‘A few considerations, generally applics- ‘ble, might be noted’ here. (a) For the ‘same system voltage, an effectively {grounded system permits use of a lower lightning arcester voltage level then the isolated system equipped with « ground- fault neutralizer. Thus, system protec: tion can be improved, system insulation can be reduced, or some of each benefit obtained. (b) When the neutral is solidly and permanently grounded, the major insulation of the neutral ‘end of the transformers can be reduced. (c) If the breakers of a grounded system are set for fan appreciable time delay, some of the faults may burn themselves out, without causing shutdown. If the breakers ere set for a short time delay, they may putting out the are, and then immedi reclose, giving only a momentary ruption quite often seceptable with the concepts of good service. So @ line-to- ground fault does not necessarily mean a long duration line outage. Stage tone peeknllaes Aah 8 Ble of See ee eee iad Sis Sporntt inthe ibe Prem 18 pp oI of ad serene ete wil be found here he 2 cebu asa ‘ad Bower Mpatice owes Ta‘HE variety of possible unsymmetrical banks are a legion. As the more im portant cases have already been covered in various publications available to the reader,* we propose here, in the spirit of ‘our title, to explain how such cases are analyzed, Y-¥ with Unequal Units, Isolated Neutral (Figure 1) If the exciting currents of the three units ate not alike because of differences fin size or other reasons, the neutral potential will not be central to the three terminals. This does not affect the load (ine) voltages, but it does affect the core losses of the anits. The smaller units will take up more of the voltage, and the neutral will move towards the Tine ter- ‘minal of the largest unit, or of the unit hhaving the largest exciting current at the same voltate. ‘The coil currents will be identical with the respective line currents, regardless of the sizes and reactances of the units, So he maximum permissible symmetrical three-phase load will be limited to three times the kva rating of the smallest unit If the three phase load is small, and large single-phase load has also to be furnished, say across phase A-B (Figure 1), it ean be done satisfactorily by seeing to it that C is rated at least one-third of the three-phase kva, and A and Bis each rated at least one-third of the three-phase ‘eva, plus 57.79, of the single-phase load kkva_ If the single-phase load is a lighting oad (with unity power factor), and the three-phase load is a power load with 186.6% power factor, then in unit B the ‘two component currents will be in phate with each other (dotted arrows), calling for their arithmetic addition in 'B, The ‘multiplier, $7.7%, comes from two facts: first, only’ half of the single-phase kva is furnished by B, suggesting the factor 0,50 for B's share of the single-phase leva, But as the single-phase kva load is not at twice the voltage of B but at only 1.73, times that of B, the half-load current will be 15.80% more than what it would Ihave been ot thewoltage of B. So we have the factor $7.74}. instead of 80% for the single-phase load component. PART VI The analysis of unsymmetrical banks. Af the single-phase load is three-wire, as it generally is, this isolated Y-Y bank {is not adapted to it. A mid-tapped euto- transformer, connected across lines A and B, would be needed to secure the neutral for a three-wire single-phase system. Two other alternatives would be as follows. fall the neutrals (including that of the primary source) were grounded, and a fourth wire provided on each side for the neutral currents, thret-wire single-phase loads, such as at 120/208 volts, could be furnished. Then each phase would be independent of the others, and the size of each transformer would be determined by the load connected to that phase of the circuit ‘The other lternative would be a ¥- delta connection, described in the follow= ing section, Y-Delta with Isolated New (Figure 2) Figure ? shows on isolated V-delta system, with the mid-point of one of the delta windings brought out and grounded, to furnish 120/240 volt service on that phase. Considering the distribution of ‘Unree-phase loads in such @ bank, we can tart with the recognition that a sym- ‘metrical three-phase load on the second ary lines will produce substantially PRIMARY FIOURE 1, ¥-¥ connection with wnnaust PRIMARY al fe Ic FIGURE 2, Yodete symmetrical threesphase currents on the primary lines: and, a8 the primary side fs in Y, call and line currents vill be alike. Thus, the three-phase load will be divided eubstantally equally, a in YV-¥, regardless of the sizes and react- ances of the three transformers, Con- sidering the single-phase load, et us fssume units A and B are of equel size, and unit C is larger. Whatever current flows in the secondary C will have its counterpart in the primary C, and this in its turn will have to return half ‘through A end half through B. That is the load in A or B will be one half of that in C regardless of their kva ratings and resctances, Then, depending on the relative power factors of the two loads, the load on unit C will be as high ax 32 of the three-phase va plus % of the single-phase kva; and the Toad on A or B Wil be as high as 34 of the three-phase teva plus 14 of the single-phase kv, To make this more concrete, assume a 15- kkva three phase load, and a 43-kva single- phase load. One might have wished (and believed) that this could be hendled by a bank in which AwS kya, B=S kya, and C5445 =50 va. But according to the foregcing analysis, to be suitable for any combination of power factors, A-and B will have to be $415 =20 ka ‘etch; ond C, $4+30=35 kva, SECONDARY Single; pnose tood SECONDARY Teste Sagle-phase oodPRIMARY FIGURE 3. Grounded ¥. Grounded Y-Della Bank on a Grounded System (Figure 3) If the neutral of the source supplying the ¥ is isolated, grounding the neutral of the ¥ would have no effect on the dis- tribution of the load, But if both neutral fre grounded, then it may be seen that for the single-phase load indicated in Figure 3, unit C is in parallel with A and B in series, and the division of this load between these two parallel paths will very inversely according to their leakage im- pedances. Similar comments apply to the toads on the other phases. For the last example, a 15-kva three-phate load and a 445-kva one-phase load, let us see how two Sckva units (at A and’B) and one 50-kva unit (at C) would work out, assuming ail units have the same percent reactance at their respective rated loads. Reducing the reactances to a common bate, say 50 eva, the reactances will be in the follow- ing proportions: TABLE 1 (50 bv} Als wo) BUS wa} ‘ond 8 in sores {ond Cin tre 1 and Cin coves ater ‘To simplify the example, let unity wer factor be assumed for both the single-phase and the three-phase loads, The three-phase load may be thought of as in delta, which makes it easier to see that the (4545) kva load across the ines a—>b will add up to 50 kva; and of this, 20/21 fraction (= 47.5 ka) will be fed by C. The rest of the current, repre- senting 2.5 kva, will flow in A and B in series, loading up each one to 2.5 kva. ‘The S-eva load across b—e will load A up to (11/21) 5=2.6 kva, and B and C each to 2.4 kva; and somewhat similar comments obviously apply to the load across the Fines a— ‘SECONDARY Single-phese food The orrows are for ‘Cos the relerence hose snk on @ grounded system, ‘Tabulating these component kva loads, together with their relative phase angles, we obtain the results shown in Table 2. It is seen that the bank is perfectly ‘adapted to the combination load, Comparing this with the previous case, it will be seen that if the ¥ neutral is now disconnected, the smaller units will burn cout. Different power factors would, of course, modify the above load distribu: tion, inereasing it in some places, and reducing it elsewhere, Combining the various component eur- rents or kve's properly (vectorially) in such cases is a painful matter, and even the experienced has to be wary. The scheme used above regarded each line tosline load as a single-phase Toad, and oo © 4 ay Acrows opproptiote for "A" os the single-phose reference phase Ep tec tey TABLE 2 From bee Ua) 7.5 te inephove wine 460° trom phowe C 250 rm phere A 2.6 two inophare with Phose A 25 tye Feo" trom phase 8 2 wo 66" tom phone Be 2 we 60" tron prose C 2 kee Seam the voltage of that phase as the reference vector for that transformer. From this point of view, the other two phases are "=60° (instead of 120° or 240°) from the reference phase; their projection to the reference phase is (0,5) times in both ‘cases, and the three items in the last column add arithmetically. The srrow heads in Figure 3 represent the calcula tion for unit C, and give expression to the thought thst Ey+E,=E, They don't apply to A or Bras the reference phase, ‘The arrows appropriate to unit A aé rel ference, are shown in Figure 4A, exprest- ing the thought that Ey+Ec mE, and the arrows appropriate to B as reference are shown in Figure 4B, expressing the thought that E, +E, = Ey. It is not our thought here to urge this point of view. Those who like to work three-phase vectors need not change or apologize. In checking the foregoing results by that methed, one will have to bbe sure whether to add or to subtract each component. Open Y-Open Delta (Figure 5) When the three-phase load is small and the single-phase one relatively large, the connection shown in Figure S—open Y primary and open delta secondary—is sometimes considered. If the neutral of the source is also grounded, thie is a0 operative connection, and has the ad- oe ' ¢ Ay Aerows oppropricte for “B"as the Single-phase reference phose Eatec* Ee nates WSH12412 2500. 12542641225 be, 12541242605 ba, 13vantages of few units and simple and somewhat advantageous losd division. Obviously the coil currents are the same fas the line currents on the open side, and therefore unit A~a will carry only its share of the three-phase load; wh tunit C—e will bear not only its share of the three-phase losd but also all of the ingle-phase load. Ie the delta were closed, the current in A=a would have been 57.7% of the three-phase line current, and its cor- responding kva rating would have been (06) three-phase load. But as its current is now the same as the line current, its ating will be (1/0.577) times as much, for (34) (1/0577) =57.7% of the toad, representing a premium of 15.5% in capacity. This, however, need not be con- sidered exhorbitant. The worst that can happen to unit C—c is that the three- phase ond one-phase current could hap- pen to be in phase with each other, adding arithmetically, in which ease the kva capacity required of C—e would be 0,577 xthe leva of the three-phase load-+ the kva of the single-phase load. AAs the primary neutral (the ground) will now have to carry load currents, such Transient stretses in the winding, a matter within the control of the designer; and stresses to ground that may appear at the winding terminals including the neutral, a matter considerably within the control of the ‘operator. Although both matters are of vital importance to the safety of a trans: former installation, the present discussion may be more profitable to the reader if wwe devote it to those matters over which hhe may have a meature of control. The broad field may be more conveniently discussed under two headings; namely, windings on which abnormal voltages are impressed from an external source to which they are connected; and “floating” windings, in which abnormal voltages are induced either magnetically or electro: statically, In the frst group we may con~ sider algo the transient phenomena that are peculiar to autotransformer ite Brett fete has an ate rein an ats 14 PRIMARY FiouRe 5. 1 system is suiteble only for very small Toads or for four-wice systems. All dissymmetrical banks tend to un- balance the voltage under load. Figure S isan extreme case of dissymmetry. ‘A rather important fact, not recog- nized extensively enough, is that three phase motors are generally sensitive to unbalance in voltage, their current un- balance being $-10 times as much as the voltage unbalance. ‘The reason for this is that an unbal- anced three-phase voltage amounts to two symmetrical three-phase voltages, the larger set with positive phase rotation, the smaller one with negative phase rotation, To this negative component, PART VII characteristics Windings Connected To A Line Let us frat consider the simple trans former, and then the auto-transformer, ‘The level of the abnormal voltages that ‘can become imposed on the high-voltage line terminals of @ winding connected to fa transmission Hine is obviously deter- mined by the line insulation and the lightning arrester, and needs no particular izcussion heres but the matter of neutral potentials may’ call for some comments, Unless specially designed to the con: trary, a winding hit by an impulse wave tends to vibrate electrically somewhat like @ piano wire hit by 9 keyboard hammer. The frequency of the oscillation and the relative amplitudes of the monies are determined by the design of the winding (or piano wire). Such: phe- nomena are therefore called free or char- ‘acteristic cocillations. In a free oscillation, there are “loops” and “nodes” as show in Figure 1. The loop is a point that undergoes the maxi Open ¥ prinery with open delta secondary connection the motor offers, not a generated counter ‘em£., but a leakage impedance only. As this leakage impedance is generally in the range of 10% to 20%, it follows that ‘given percent negative-rotational com- ponent of voltage will cause 10 to S times ‘as much percent short-cireit current in the motor. A 6% voltaze unbalance may sound as nothing serious, but if it results in at least 30% current unbalance in the three-phase motors, it eannot be ignored so lightly. So, in’ three-phase systems ‘subject to appreciable voltage unbalance, it would seem to be common sense to measure the unbalance in the currents of some representative three-phase motors fand avoid motor burnout from this of transformer windings. ‘mum eyelic change, between maximum positive and maximum negative values; fand the node one thet remains unchanged. In Figure 1, a, @ are nodes for vertical motion; b, « loop. But if i sidering vertical motion, consider a vertical force, a, @ are loops (points of force maxima), and b a node. And #0 it ie that when a string breaks while playing, it is apt to do so at a support. Thus a force node is « motion Toop, and vice versa. In electrical trims, in a free oscillation, a current node will ‘be a voltage loop, and vice versa, 50 that where the current is maximum, the voltage will be zero, and vice versa. As there can be no current in an isolated neutral, the voltage to ground at that point will be a maximum, As the poten- tial of the line end is fixed from the external source, the neutral end vibrating freely may assume potentials higher than that impressed atthe line end. Both theory and teste indicate that, in an os- cillation under an impulse wave, an iso- ated neutral may acquire a potential as igh as two to three times that at theDX (7) tnse for induced lightning voltages. A Nedes ond toops ore nodes ‘and Us toep for vertical movements oy @ fore loops and Blt node tor vertical force FIGURE 1. Diegrom of winding evita line terminal.! Evidently the isolated neutral bushing and the isolated neutral tend of the winding may need more insu: lation strength to ground than docs the Tine bushing and the line end of the winding, unless by appropriate design the initial voltage distribution is sub- stantially Hnearized and the winding is rendered non-oscillating or substantially 80. As is well-known, grounding the “solidly” (without impedance) yermanently” permits substantial ‘eduction in the insulation of the neutral fend and a substantial saving in the cost of the transformer. The foregoing comments apply in a general way to all winding connections, bout it may be of some interest to compare the isolated Y-Y connection of single- phase units, Figure 2A with the Y- Delta, Figure 2B, and the Delta-Y, Figure 2C. Asan oscillation involves both in- ductance and capacitance, the lower the inductance, the higher will be the result. ing frequency and damping, Little differ. ences, such as that between a 6 percent and a 12 percent reactance may not be significant, but that between the leakage reactance (of eay 10 percent) and the ‘magnetizing reactance (of say 10,000 per- cent, representing 1 percent exciting current) could be significant. Naturally, the power loss associated with the in: ductance~or more precisely, the power factor of the inductance—is even more important as the major damping agent for the oscillation. Dielectric losses come in too, but generally they are much smaller than the I'R losses and the mag- netic losses. Figure 2 illusteates, ina simplified ‘manner, whether it is the magnetizing reactance or the leakage reactance that centers into the fundamental mode of oscillation of the winding. The capaci- tances of the primary (high-voltage) windings to ground are shown as massed at the middle of each phase, The line- ‘end half of the winding in series with the ei s tea reg Te i ate PS capacitance to ground constitutes an oscillatory circuit. It is assumed here for simplicity that the incoming wave is identical in all theee lines. This is often direct stroke may be in only one line, but its analysis is somewhat more com: plicated, Tt will be recalled from a pre- vious installment that the symmetrical ‘casein phase in all three lines—con- stitutes @ zero-phase-sequence set, and will therefore encounter the sero-phase- sequence reactance of the windings, Tn Figure 2A no induced secondary (L. V.) currents can flow to neutralize the magnetizing effect of the primary ‘currents, seeing that the secondary neu: tral js isolated. By virtue of the lowness of the voltage the capacitance currents are negligible. So the windings offer a core-magnetizing reactance to these cur rents, This high reactance is the zero- sequence reactance of Y-Y connected transformers. But in Figure 2B, the currents induced in the delta-connected ‘windings can circulete within the delta, and so the windings offer oaly the leakage reactance between the primary and the PRIMARY HIGH VOLTAGE secondary windings. This low reactance is the zero-sequence reactance of Y- delta connected units As we have seen before, the zero: phase-sequence reactance of three-legged three-phase transformers is intermediate between leakage reactance and magne- tizing reactance. Figure 2C shows @ delta high-voltage, Y low-voltage, bank. Each phase of the delta may be said to have a neutral of its own-—its mid-point—as it will be noted that in each phase there are two currents flowing from the lines towards the mid-point of that phase. These two currents neutralize each other's. mag- netizing effect on the core and encounter only the leakage reactance between the two halves of each primary winding, Thus, there is a large difference between the impedances of delta-connected. windin and isolated-neutral Y-Y windings with respect to their fundamental modes of oscillation, though both banks are iso: lated. Let us consider now some transient phenomena peculiar to autotransformers, SEconoaey Low vowtace FIOURE 2A. The ¥-¥ connection of «sin MoH vourAse abate unit with the secondary Low vourace, SN oe the dete windingi Jsolated-neutral ¥ Autotransformer ‘The series winding (S, in Figure 3) of : an autotransformer is an inductive series link between the incoming line A and the outgoing line B. Therefore, these is @ strong tendency for an incoming surge 1 to become concentrated across the ob structive impedance of S. Comparing Figure 3A (no delta) with 3B (with ter tinry delta), the ratio of the two im pedanees may be seen to be that of a magnetizing reactance to leakage re | actance, which may be at least 100: leaving little need to resort to advanced mathematica to appraise the lative voltage concentrations, at least quali tatively? ‘Another important coviideration is that in Figure 3A the voltage developed across $ will be alepped up into P approx. imately by turn ratio, raising the po. tential of ¥ to a dizzy height, In a tap: changing voltage regulator with, say, one i Percent steps, the ratio of common to i series turns on the first step will be 100:1, f and 20 the seriousness of this overvoitage : at the neutral may be appreciated. In Figure 3B (with a tertiary delta), not I only the voltage across S will be a small | fraction of that in 3A, but ‘in general [ | only a fraction of the reduced voltage across S may be induced in P, due to the magnetic shielding action of the induced Circulating current in the delta, ‘As any series inductive device, even fone of a reasonably low 60-cycle react. ance, is likely to develop considerable voltage when subjected to a 10,000~ i 50,000 ampere surge current wave, if is i generally wise to protect the series wins ings of autotransformers with thyrite » resistances. This is particularly desirable in step voltage regulators in which, not only the insulation safety of the winding Parts must be considered, but that of the ijt eltong nt ale has tobe Calf eat ei ” sSoLATED FIOURE 3A. Y-Avtetenatormer ithe | 6 Grounded-neutra! Autotrantformer We have seen in the earlier issues that ‘2 grounded system, the grounded neutral autotransformer ig a highly icheine of operation, especial: is also @ delta winding in the bank. While a solidly grounded ‘neutral is safe regardless of any series-winding reactance considerations, it may be seen in Figure 4 that, because currents can now flow freely in the neutral, the re- actance of the series winding to surges willbe very low: in the absence of a delta, it will be only the leakage reactence be. tween S and P, a low Sgure; and in the resence of a delta, still les, In this case, the delta is not essential for surge con. siderations, but it will keep the harmonic currents cut of the neutral: and, of ‘course, it is also a backup protection in case the circuit between the two neutrals should become temporarily opened, as by fine breakers opening on one side. Another precautionary comment that may be made here is that, as delta wind- ings in autotransformers are generally: designed for 35 percent of the physical capacity of the series or common wind- ings, it i wise to make certain that, under the mest unfavorable system fault con. mn, the delta winding is not overs loaded. Floating Windings? A “floating” winding is a winding which is isolated from ground and all external circuits of appreciable capaci. tance. The electrostatic potential of such @ winding is determined by the capaci« ‘tance distribution of the transformer, ‘and under certain conditions indicated below, such a winding may be subjected to high potentials to ground by electro static induction from the high-voltage winding, A floating high-voltage winding may also be subjected to high voltages by electromagnetic induction from the low TE aitn fem patinion on an08 op ‘he Cahir eer Com fs sovareo FIOURE 98. Y-Avtotrem voltage winding. ‘The fact that the major insulation of & floating winding may be safe in a given case does not guarantee the safety of the Dushing and connected idle bus bars or ‘other equipment of low capacitance to ‘round. Therefore, itis advisable to take Precautions to prevent over-voltages and resulting sparking from such parts to Ground. ‘The general nature of these troublesome conditions and the recom: mended methods of protection are as follows. Electrostatic Induction Electrostatic induction is hazardous when the isolated winding is. ata much lower voltage rating than the inducin winding, because this type of induction is determined not by turn ratio but by the relative distribution of the capacitances, ‘among the windings and ground. Gener. ally the capacitance distribution teads to ‘much higher potentials in the low-voltage windings than turn ratio would have done. In @ single single-phase transformer installation, with the high-voltage ter. minals connected to two lines of a three. phase system, the high-voltage winding ormelly has an a-c potential to ground (averaged along the winding) approxi: mately equal to one half of the linesto: ground potential, This will tend to induce in a floating low-voltage winding a cor. fesponding electrostatic potential based ‘on the capscitance network of the winde ings and ground, Likewise, high transient potentials on the lines, that raise the average potential of the high-voltage ing, will induce high potentials in a foating low-voltage winding, possibly intensified by internal oscillations. To ilusteate: in commercial impulse testing of ‘high-voltage windings, if the lowe voltage windings were allowed to fost, the low-voltage bushings would flask over to ground in most eases. Therefore, it is standard practice to connect the Tow.vottage bushings to ground through a resistance during that test, In symmetrical three-phase installa: tions, the vectorial average of the normal frequency potentials of the three phases to ground is zero, and therefore the cor. responding average normal-frequency po- tential to ground of a symmetrical threes hase floating winding also will be sero, Aa the clectrostatically induced potentials im the three secondary phases tend to Reatralize each other by an exchange of small capacitance currents, and at the operating frequencies the windings offer Aegligible impedance to the flow of these kurrenta, the maximum focal values of the clectrostatically induced" norma frequency potentials will be negligible, However if for any reaton the neuteal of the high-voltage winding should shift, oFthe potential of the high-voltage windings should be raised ag a whole, a corre sponding electrostatic potential will bein duced in 2 floating winding on the trans: formers, This is likely to occur under the following conditions. 1. Lightning |, Lightaing on the high-voltage lines is ‘the source of maximum electrostatic induction from the high-votta to a floating low.voltage windi the steepness of these potentials, some or raised to much higher potent average of that winding. The worst case for electrostatic induction to a floating winding ia when lightning raises the potentials of all three high-voltage lines equally. 2. Suitehing with Single-pote Switches or Blowing of Fuses In switching with single-pole air-break switches, of in the blowing of two fuses, fone line may be closed and the other two appreciable length of time closed line inducing an sbnormal potential in a floating low-voltage winding, These ‘voltages are fers than those in case (1) but are of relatively long duration 4. Switehing by Oil Circuit Breaker ‘As the three contacts of an oil circuit breaker cannot be expected to open or close exactly simultaneously, their op- eration also leads to the same situation at in (2) but with very much shortened duration, 4. Line-to-ground Fault on Tsolated Three-phase System A Iine-to-ground fault on an isolated three-phase system shifts the neutral and average potential of the high-voltage windings and induces a corresponding abnormal electrostatic potent floating winding. 5. Dynamic Overvotioges Unbalanced fault conditions, even in a rounded-neutral system, may cause neutral shift ofthe high-voltage windings ‘and. electrostatically induce a corre: sponding abnormal potential in a foeting winding. 6 Other Cases Accumulation of electrostatic charges and other obscure causes may lead to ab- normal potentials in Rosting windings. Electromagnetic Induction While electrostatic induction can pro- duce dangerous veltages when the float- ing winding is of a much lower voltage rating than the inducing winding, electro- ‘magnetic induction may produce danger- fous voltages when the floating wi ‘of much higher voltage rating than the —. fs IHOURE 4A, Groundedneiraloletransformer inducing: winding. An impulse voltage concentrated across 2 low-voltage wind. ing er portion thercof may be stepped up roughly by turn ratio in a floating high- voltage winding; ond if the factor of safety of the low-voltage winding for its ‘own voltage classification is much higher than that of the floating high-voltage ‘winding for its classification, then the hazard will be greater in the high:voltage winding, Furthermore, the stress in the high-voltage winding may also become complicated and intensified by the ‘oscillation cceasioned by 2 steepsfront indaced voltage. Recommendations ‘When equipment is exposed to danger cousoverveltages, its generally customary to provide suitable protection in the form of lightning arresters for the unit, If the ‘conditions discussed sbove apply to a floating winding, provision for its protec- tion may be advisable depending on its ‘connection. 1. Windings in Y Connection (@) Directly ground the neutral and provide lightning arresters on the three Tine terminals. Low-frequency electro: static induction will be taken care of by the grounding of the neutral alone, but induction due to impulses or switching ‘surges makes arresters on the line termi nals advisable, (b) When for some reason the neutral may not be directly grounded, connect a lightning atrester between the neutral ‘and ground, in addition to the arresters ‘on the line terminals, 2, Windings in Delta Connection (2) Ground one terminal of the delta, ‘ond connect lightning arresters between ‘each one of the other two terminals and ground, (b) If the grounding of one terminal of the delta is objectionable because of large fault currents, the grounding may be inacle through an approprigte resist: ance to limit these currents to the desired FIGURE 4B, Groundad.asuteal withthe det value. In this case a lightning arrester should be connected between that termi- rnal and ground, the same as for the un- grounded terminals. (c) Tn systems with protective relays responsive to electrostatic unbalancing of the lines, the grounding of a terminal of the delta, even through a resistance, may be objectionable, In such cases grounding ‘may be omitted and an arrester connected between cach delta terminal and ground. ‘A practical difference between this case () and cases (a) and (b) is that in cases (@) and (b) the cecasions for the arresters to discharge will be greatly reduced. 3, Selection of Arresters Im selecting arresters for application between the line end of floating windings and ground, an arrester rating must be chosen which exceeds the maximum ‘potoer frequency voltage which can be im- pressed across the arrester in case one of the terminals is cither temporarily or permanently grounded, ‘This voltage is a function of the con- nected eystem parameters, as well as the transformer charecteristics; and, there. fore, cach application must be indi- vidually examined. General guidelines, however, indicate: 1. For delta-connected or wyecon- nected, ungrounded windings, the arresters are generally in the order of 100% of maximam Tine-to-tine operating voltage. 2, Ifthe winding is wye-connected and the neutral solidly grounded, the arresters may be in the range of 12% to 80% of maximum line-to- line system operating voltage. 3, Ifthe winding is wye-connected and a neutral arrester is used, the arrester rating may be approx mately 70% of maximum line-to- line aystem operating voltage. Tt should be stressed that specific ays- tem conditions and parameters may require arrester ratings in excess of those indicated. v7PART VIII Calculation of neutral shift of isolated-neutral autotransformers \RRBFORE an cutotransformer is oper- ‘ated with neutral isolated, the neu- teal shift, during ground faults on the aystem, should be throughly investigated to make certain that the autotransformer neutral insulation and neutral lightning arrester, if one is used, will not be sub- jected to excessive power-frequency volt- ‘ges. This section deals with methods of calculating neutral shift. As there are likely to be a number of “neutrals ina polyphase system, it might be worthwhile to clarify them first. Figure 1 shows a power source, S, in Y connection with its neutral point, Nyy grounded through an impedance Zy. One neutral is Ng; ariother ia ground, G, whose potential ie. logical reference potential ‘28 "neuteal” or zero potential at ll times. ‘An autotransformer, , is shown with its ‘own neutral point, Np, which may be at still another potential. Then there Is such 2 thing as the “neutral” of the lines (not shown in Fig. 1) representing the center of gravity of the line potentials and nor- mally coinciding with G. The potential of thie neutral is not necessarily zero with respect to ground under all conditions. It can be identified as follows: Connect three duplicate resistors (or impedances) in laolated ¥ across the lines. The potential of the neutral of this bank will be the cen- ter of gravity of the line potentials, and it may be called the "“neutral of the lines.” ‘The suggested bank of impedances to identify this neutral may be capacitors, in which case the capacitance elements may be the capacitances of the lines them selves. If the capacitances are not per Feetly symmetrical, the “neutral” which they identify will be different from that which a symmetrical impedance bank ‘would identify. In such a ease, the neutral of the lines will have two different mean- ings. Moreover, the neutral of the lines may be different at different points of the rystem, such az on the primary and secondary sides of a transformer. Let us calculate the potentials of some ‘of these “neutrals” when one of the lines is faulted to ground. The method of sym- metrical components is so convienent for such problems that we cannot refrain from Care 1. Source Grounded through on tmn= pedance, Aviotransformer Isolated, Step- ping Down, Fault on Primary (HY) Lint Figure 1 shows the diagram of connee- tions with a fault to ground on line A be- tween the source and the autotransformer. Figure 2shows the positive-,negative-,and rero-tequence equivalent networks inter- connected to represent the assumed fault condition, The various quantities shown in Fig. 2 are defined as follows: Zags Zyq and Zug are respectively the per-unit positive, negative, and zero- sequence impedances of the source 8. Zqx is the per unit zero-sequence im- pedance of the neutral impedance. Tt obtained by taking the ratio of 3 2yy (chins) to the base impedance (chms) at ‘the terminals of the source. Zyy it the per unit positive-sequence impedance of the autotransformer, T. Zein Za. and Zq are the per unit branch impedances of ‘the 3-terminal zero {quence equivalent circuit which represents Flouae 2, rece the zero-sequence impedance character- {sticg of the autotransformer with neutral grounded. The ideal autotransformer, R, shown in the zero-sequence equivalent network takes care of neutral isolation® All per unit impedances are based on rated linesto-neuteal voltages and the seme kvs, Ris an ideal autotransformer, the ratio of transformation of which is exactly the same as that of the actual autotransform: cr being represented. Tt should be noted that the low-voltage side ofthe ideal auto- transformer is connected to the H-side of the equivalent network and the high- voltage side of the ideal autotransformer is connected to the L-side of the equiva- lent network. This might seem to be op- posite to what it should be, but it is cor- eet. yg is the per unit neutral shift of the source neutral, Ng. It Is based on rated line-to-nentral’ voltage at the terminals of the source. Eqq is the per unit neutral shift of the high-voltage lines. It is based on rated e-to-neutral voltage of lines A,B and C, Eye is the per unit neutral shift of the autotransformer neutral, Nr. It is based fon the rated line-to-neutral voltage at the high-voltage terminals of the autotrans- former. Eq is the per unit neutral shift of the low-voltage lines. It is based on the rated line-to-neutral voltage of lines a, b and c. All ofthe neutral shifts are with respect to ground potential, See Re cette Mie tacaathene Poni pp itesFHOURE 3. seurer grounds aval Hetoted foul! With rated positive-sequence source voltage applied, it is obvious from the interconnected ‘equivalent networks of Fig. 2, that Zon Ene Flim tea lon . and 2a tes Bun Eut~ 7 lq tlgtin Ey." Eo ° Eu Bi Base mation of idee autotransformer where r=: ‘=ratio of transfor: ‘Thus we have simple formulas for cal culating the chifts of the four “neutrals” lunder the specified fault condition. It should be noted that under this condition there is no fault current in the autotrans former, ‘The autotransformer neutral, Nz, has shifted with respect to ground, but not with respect to the line potentials. The source neutral hes shifted with respect to both the ground and the lines. Because there is no fault current in the autotransformer, the values of Zu Zu, ‘and Ze do not affect the results, Hence, under the assumed fault condition the presence or absence of a delta-connected winding does not affect the shift of the autotransformer neutral, Step-up or step-down connection of the autotransformer also is immaterial to the Value of the neutral shift with respect to the primary lines, but agiven neutral shift may be very differently appraised from the standpoint of the safety on the low- voltage system. To illustrate this, let the neutral shift given by equation (2) be 0.50 (of the high-voltage line-to-neutrel voltage). Let the low-voltage rating of the autotransformer be one tenth of that of Figure 4. the high voltage. As the high- and low- voltage neutrals of the autotransformer are one and the same point, the 50% shift of this neutral in terms of the high-side voltage will be ten times as much, or 500% shift, for the low-voltage system. This is an accentuated illustration of ‘the reason for the maxim that'a proposal for the use of an isolated-neutral auto- transformer with more than 2:1 ratio should be carefully scrutinized, because with increase inthis ratio the economy of the autotransformer over the conventional transformer gets smaller and such troubles greater, Fe Mel tetete pete i a tng jotentias of touree tines ond vorlevs foul tn secondary sige. ose On Secondary Side of Avlotransformer, Same As Case | Except Fault ‘The connections for this case are shown, in Fig, 3; and the positive, negatives, and. zero-sequence networks interconnected to represent the assume fault condi shown in Fig, 4, For this case no simple ‘expressions such as equations (1), (2) and (3) are obtainable because the circuit of Fig. 4 is more complicated than that of Fig. 2. Therefore, a numerical example will be used to illustrate the method of caleulatiAssume the following per unit values ‘of impedances besed on $0,000 Kva: Zig =2ig nag) 0.0250 Zu =} 0.0100 2qu=5 0.0100 Zn, =) 0.0034 Zag 5 0.3166 ‘The autotransformer is rated 50,000 Kya, 138 to 115 Kv ond has a delta-con- nected stabilising winding. The ideal autotransformer shown in the zero:se- quence equivalent network has the same ratio of transformation as the actual autotransformer being represented, that ig re 138 to 115~1.2 to 1. ‘The impedance of the zero-sequence ‘equivalent network can be found to be Zyj 0.0809 ‘The total impedance of the intercon- nected networks is Z=2_ (j0.0250+45 0.0167) +) 0.0809 = j 0.1643, ‘Therefore, the positive-, negative-, and zero-sequence components of fault cur rent are hahaty 1.000, F0.1643, 56.08 per unit. Since Ty it (J 6.08) the Ideal auto- transformer forees Iq to be Tuy = 1.2 (—j 6.08) = ~5 7.296 per unit. imilacly Yq)=0.2 (~j 6.08) = —j 1.216 per wi Using these per unit currents, through the per unit impedances, the various neutral shifts are found to be ys =~ 0.0100) (~j 7.296) 07298 Eqn = ~G 0.0950) (—j 7.296) 2554 Eqs = —( 0.0450) (~j 7.296) 1.216) (j 0.3166) By, = ~() 0.0809) (—j 6.08) = ~0.4019 Converting these per unit quantities to voltages yg = ~0.07296%138//S=—S.81 Kv Eqn = 0.2554 138/VF= 2035 Kv Eyer -0.7133 x138/ V5 Buy -04919x115/ V5 = ~32.60 Kv =56.80 Kv ‘The phasor diagram showing the poten- tials of the various “neutrals” with respect to ground is shown in Fig. 5. The solid lines and unprimed symbols are for tunfaulted conditions: dash lines and primed symbols are for the fault con ton, Before the fault occurs all “neutral are at the potential of ground, G. If the delta-connected stabilizing wind. ing had been omitted from the autotrans- former, the values of autotransformer equivalent circuit impedances (assuming S:leg -core-type construction) would ‘change to something in the order of: Zyy =~} 9.0913 Zy, 95 0.1080 2q =} 3.5880 Using these values, the shift of the autotransformer neutral is calculated to be 204 ky with respect to ground, which is approximately 1.5 times the rated line- torine voltage on the high-voltage side. ‘The other neutral shifts are also different from the corresponding figures calculated for the autotransformer with delta-con- nected stabilizing winding, POWER TRANSFORMER DEPARTMENT GENERAL @® ELECTRIC PITTSFIELD, MASS. Ger-aan0m (SUPERSEDES GET-396A AND OEA-2605) ne Gio se ay