H ow o w t o de d e al al w i t h st at o r c o re damage RichardNailen Electrical Apparatus Jun 30, 2006 20:00 EDT
"RUSSIA IRON," "STOVE-PIPE IRON" OR JUST plain "sheet iron" -such were the materials used to make laminated cores for electromagnetic apparatus before development of the "electrical steels" we know today, with their carefully controlled additives of silicon, aluminum, cobalt, or other performance-enhancing elements. Hence the term iron loss remains in common use. However, "core loss" better describes the energy wasted in magnetizing and demagnetizing a core assembly. Core loss takes only two forms. One is hysteresis. That's the energy required to align and realign the molecular structure of steel with the movement or change in the magnetic field passing through the material. Hysteresis loss is determined by the metallurgy of the steel itself. The punching process creates internal stress within the material that can increase this loss. That's why laminations (or assembled cores) are often annealed, to relieve that "punching stress." In the 1980's, an English manufacturer of large rotating machines reported that tests "suggested" core loss around teeth and slots exceeded those in unpunched steel laminations by about 10% because of "work hardening" caused by the punching process. The second type of loss is an I^sup 2^R phenomenon, created by the flow of eddy currents through the steel's electrical resistance, driven by voltage that the magnetic field induces within the laminations. The purpose of laminating the core, with some type of insulation coating between layers, is to minimize that current flow. Both hysteresis and eddy current losses increase with the frequency of field alternation. Both also depend upon the magnetic characteristics of the material. However, another major influence on the eddy current loss alone is the electrical resistance to current flow between adjacent laminations. That so-called interlaminar resistance in turn depends upon three variables: 1) the resistivity of the insulating coating coating or coreplate on the lamination surfaces; 2)
the nature and location of structural "bridges" or contact areas that short-circuit the coreplate; and 3) the tightness with which the laminations are pressed together in the core stack. In an a-c machine stator, these two losses occur in three places: * Below the slots, in the "core," back iron, or yoke. * In the teeth, between slots. * At the inner surface or stator bore. The losses at the stator bore, involving tooth surfaces, are variously described as pole face, surface pulsation, or zigzag losses. They result from higher-frequency harmonic fluxes that interlink stator and rotor across the air gap, influenced by the respective surface discontinuities caused by the slotting. In many small motors, bridged rotor slots eliminate the discontinuity on one side of the gap. Stray core loss The nature of the pole face loss sometimes leads to its being described as a "stray core loss." Strictly speaking, stray load losses are functions of rotor current, so they do not appear at noload. The often-used formula for calculating surface pulsation loss is shown in the box on the following page. This loss is not directly measurable. It does not exist unless the rotor is in place. However, total measured "core loss" in a machine is about double the sum of the calculated tooth plus back iron losses (the multiplier ranges from 1.75 to 2.2). The division of core loss into its two components is important because, as is not always clearly understood, most causes of core damage have no effect on hysteresis loss. One exception is a winding fault that produces local overheating and arcing in the slots, sufficient to vaporize copper and melt steel. Otherwise, steel metallurgy is unaffected either by temperatures below about 1,500 If such damaged areas are small enough, as in Figure 1, repair may be possible by carefully grinding or drilling away the fused material. Otherwise, the core must either be scrapped or restacked with enough new laminations to restore the original structure. Other sources of core damage will affect only eddy current loss. One is the well-known "stripping oven" over temperature, in which the core is heated enough to char or soften insulation in the existing winding to permit its removal and replacement. Depending upon the core configuration and the oven operation, the excessive heat may destroy the inter laminar insulation throughout the core, or only in certain areas (the teeth being most vulnerable).
If damage is severe enough, increased core loss and local heating will jeopardize the replacement winding insulation as well as seriously decreasing machine efficiency (particularly at light load). Another source of stator core damage is the rotor "rub" resulting from a bearing failure or magnetic instability (Figure 2). Friction between the spinning rotor surface and the stator ID may not be sufficient to thermally damage the winding itself, but the grinding action will "smear" together the surfaces of lamination teeth, with consequent large increase in surface pulsation loss. Machining the stator bore to "clean up" such smearing can not only be ineffective, it can make matters worse. Such smearing is usually detected by visual inspection. If the line separating one lamination edge from the next is clearly visible, separation is probably adequate. If only a single, unbroken mirror-like surface can be seen, however, inter laminar contact is sure to be troublesome. That can sometimes be corrected by etching the stator bore surface with a mild phosphoric acid solution. Done either at the factory or during rebuilding, slot filing will also increase eddy current loss. The problem should not occur in the repair shop unless the core has been taken apart and restacked without sufficient care to maintain slot alignment. Once a core is assembled with excessive "stagger" or misalignment of slots from one lamination layer to the next, proper fit of coils in the slots becomes impossible. The only remedy is to broach or file the slots to remove the high spots. That creates the same kind of "smear" just described. Core tests Establishing the existence of core defects except right at some external surface can be difficult. Proponents of the "El Cid" test developed for that purpose have compared several diagnostic approaches as shown in the table at right. But even the best method should be considered only as an adjunct or follow-up to a ring or loop test (as detailed in the Appendix to IEEE Standard 56). To insulate laminations from one another, several different coreplate insulation materials have been used by the manufacturers of lamination steel. For many years, "C-3" was most common. This is an organic varnish material, easily destroyed by temperature above 600°F to 700°F. More common today is "C-5," a largely inorganic material containing "ceramic fillers or film-forming components," that will withstand annealing temperatures. It will not normally be damaged in a stripping oven. A "C-4" coating is formed by chemically treating or
phosphating the steel surface. A more recently developed "C-6" material is a silicate-resin compound for semi-processed lamination steel, capable of withstanding 1,500°F with postannealing resistance superior to C-5. Unfortunately, manufacturing records do not necessarily indicate which coreplate coating may be present in a particular motor. Visual inspection is useless in making that determination. Nor can surface appearance of any lamination reveal the quality (or even the presence) of suitable coreplate. For many years, the only standardized procedure has been American Society for Testing & Materials No. A-717, known as the Franklin test. The Franklin test involves clamping lamination material samples under a set of electrodes imposing 300 pounds per square inch pressure. A standard test voltage is then applied to the electrodes and the resulting current measured. From that measurement, the resistance across one surface of the material can be calculated in "ohms per square." Unit size of the square is immaterial, because any dimensional change increases both square width (which lowers resistance) and square depth (which raises it by the same amount). Note that this procedure doesn't directly measure the interlaminar resistance between surfaces of two laminations within a core, at whatever the core clamping pressure may be. Hence the adoption in 1996 of ASTM A-937, titled "Standard Test Method for Determining Interlaminar Resistance of Insulating Coatings Using Two Adjacent Test Surfaces," which does involve such a direct measurement. For any machine, using allowances for variation that have been determined through experience, motor designers and manufacturers can calculate losses to be expected in the first two of the three locations mentioned earlier (core and teeth). Various tests (including the "Epstein" method for hysteresis) support the calculations involved. Subsequently, total tooth plus core loss in the assembled machine can easily be checked by the standard no-load test procedure of IEEE Standard 112. However, judging the accuracy of the calculation procedure requires some means of separating the hysteresis and eddy current components. That's done by measuring total core loss at two different frequencies, such as 50 and 60 Hz. Suppose total measured loss in watts is P^sub 1^ at 50 Hz, and P^sub 2^ at 60 Hz. Then, the equation in the box at left applies. If such testing isn't feasible, a common assumption is that hysteresis and eddy current losses are equal. However, the relationship can vary considerably. Typical values of the hysteresis-toeddy ratio have ranged from 0.889 to 1.8 in 449 frame two-pole machines, for example.
Mechanical considerations What are the mechanical features of stator core construction that can influence eddy current loss? The most important one is the burr formed by the punching or stamping process at all cut edges in each lamination. (see Figure 3.) As dies wear during production, that burr will increase in height. A common practice of large motor manufacturers has been burr grinding of laminations. After punching, each lamination is passed through a belt sanding station that smoothes off the burrs. Coreplate must then be added later to cover all bare metal areas left by the sanding. Such deburring has been found to reduce total core loss by 4% or 5%. Subsequently, care is normally taken to stack all laminations with the slot punching burrs in the same direction. That's important because the region at the tooth tip is the most sensitive to interlaminar current flow in general and to "surface pulsation" in particular. The sharp edges of the burrs on adjacent laminations will cause good electrical contact if opposed, which is minimized if they "nest "together-see Figure 4. Laser-cut laminations are not subject to this precaution. However, cost and production time constraints still support punch-and-die manufacture of most laminations, particularly for mass-produced ratings. Another influence on eddy current loss is core stacking pressure. Interlaminar resistance will decrease, driving up the eddy current loss, as that pressure is increased (Figure 5). Making sure the stack is compressed tightly-yet not too tightly-is a challenge for any manufacturer or rebuilder. The problem area is around the teeth. Despite the use of many different types of "fingers" to clamp teeth together, some flaring or splaying is inevitable. One reason is that laminations are seldom perfectly flat or of uniform thickness throughout. The dimensional variations involved here are extremely small, and seldom accurately measurable outside the laboratory. Nevertheless, the buildup over a long core stack can be considerable with typically 51 laminations per inch of length. The surface oxide coating on the steel could reach 0.00009 inch. The coreplate coating thickness may be taken as 0.00022 inch (total for both sides). Add 0.0004 inch for burrs, and another 0.0001 for deviation in flatness. All that makes a total possible addition per lamination of 0.000729 inch. The usual result of all these variations is a finished stack length longer at the ID than at the OD, by 1/8 to 3/8 inch. Consequently, what the retained pressure will be throughout the stack
can never be known. Subject as they are to continuous vibratory forces in the axial direction at twice line frequency, teeth not tightly restrained can abrade coil insulation, or eventually break loose to cause more serious damage. Deciding whether the finished core is "tight enough" is necessarily somewhat subjective. One criterion is the "knife test," in which the attempt is made to insert a knife blade between laminations, and the degree of core tightness is judged by how deeply the blade may penetrate. Obvious drawbacks to such a test are: * What should be the blade thickness? * What depth of penetration determines that the stack is too loose? * Should the test be made at the O.D. or the I.D. of the core? Each repair shop must determine the appropriate answers. The stator core restacking specification issued by a large public utility describes a similar test this way: "It should not be possible to insert a 0.020 inch feeler gauge between any laminations." (Although it may seem unrealistic, that at least answers questions 1 and 2.) Rebuilding cores The first step in core building or rebuilding, however, is to decide on a stacking pressure expected to produce a tight stack. The second step, for long cores in large machines (and "long" has its own subjective definition) is to specify appropriate "intermediate pressing." To align the laminations properly, any core must be stacked on some kind of supporting structure -a mandrel fitting against lamination ID or OD; guide bars or "drifts" in certain slots; or keys or studs on the OD. As laminations are stacked up, increasing friction against those supports will resist the clamping pressure. The result is variation of core tightness throughout the stack length, from a maximum near the top of the stack to minimum near the bottom. To equalize the compaction throughout, clamping pressure is applied to the first quarter, third, or half of the total stack. The next section is then added, pressure applied again, and so on until the finished height is reached. Pressing force often ranges between 100 and 150 pounds per square inch of gross lamination surface area. No standard applies. The area is generally taken as the overall difference between OD and ID, without regard to the material removed by slotting. Some manufacturers, however, consider only the area below the slots.
The definitive study of core pressing is a 1964 paper presented to the British Institution of Electrical Engineers, titled "Pressing and Clamping Laminated Cores." Its authors experimentally developed the relationships shown in Figures 6 and 7. Their conclusion was that a stacking pressure of only 40 psi ensures adequate core tightness, provided that such pressure is maintained throughout the core for its entire life. Probably the most uncertain condition in stator core building is the effect on lamination short-circuiting of the manner in which the pressed core is held together. Segmented cores are stacked on studs or dovetail key bars, with relatively poor electrical contact along the OD (Figure 8). Smaller, one-piece laminations may be held together by welding or dealing (Figure 9). Certainly the most solidly connected laminations will be in a core that is welded together on the OD In small sizes, weld beads are deposited directly on the laminations. One difficulty is that the heat involved can cause undesirable metallurgical changes in the lamination material (a TIG process is recommended). Also, a good bond is difficult to achieve with silicon steel. Larger cores will be welded against supporting steel bars to which clamping end plates are attached. Limited experiments have shown that core losses do vary with the pattern of such welding, but the range is neither great nor readily predictable. Theoretically, for any machine, the greater the back iron depth in the core, the farther the OD is removed from the most intensely magnetized region closer to the winding, and the less will be the influence of interlaminar contact. In one 500 hp 3,600 RPM machine, tests of a "fully welded" core showed 22% more core loss than in an unwelded version; that amounted to less than two-tenths of a percentage point drop in efficiency. However, the result was not conclusive because of other differences in core construction that also influenced losses. In contrast, two 350 hp 720 RPM stators built with widely different amounts of welding exhibited identical core losses and temperature rises (see Figures 10 and 11). When unacceptably high core loss is observed in a small or medium motor, a simple remedy may be what some have called watt knocking, thumping, or slamming. In its simplest form, this involves striking the unwound core sharply with a hammer, or pounding it a time or two against a hard surface. The mechanical shock tends to break apart interlaminar "sticking" caused by core annealing or machining. A more sophisticated and better-controlled process applies the impact of an air-powered vibrating hammer for a minute or so.
Large cores require more elaborate methods of repair. Before deciding what to do, you should study a core carefully to judge the nature and extent of damage. Loop or ring testing, with infrared or other measurement of temperature on all accessible core surfaces, is the only way to do that. The "El Cid" test, employing the same basic principle, is suited to the largest machine sizes and will highlight damaged areas well away from external core surfaces. In a loop test to evaluate stator core condition during the repair process, both hysteresis and eddy current losses are combined in the result. Two serious limitations affect the measurement. First, only the back iron or core region is fully magnetized. Second, and more important, those losses will reasonably match those in the assembled machine only if the magnetic field strength during the test matches the design value. Assuming a "typical" figure cannot be expected to produce losses at the operating level (assumed core flux density for years has been in the range of 60 to 100 kilolines per square inch-about 1.5 Tesla. Yet motors of all sizes today often use actual back iron flux densities of 110 to 130 KL/square inch). The proper indication of core condition is not the total temperature observed after some period of time, but the rate at which temperature rises. Readings need to be taken at frequent intervals as the test proceeds (for at least 20 minutes; some recommend 40 to 60 minutes). One good rule of thumb: temperature in a "good" core should rise no more than 5°C to 10°C within half an hour, whereas in a bad core the rise can be 15°C to 20°C higher. If damage is extensive, the difference may be much greater-see Figure 12. In a large motor (at least 1,000 hp at 3,600 RPM down to 400 hp at 514 RPM) economics will usually dictate restacking a damaged core. Depending upon its construction and the location of the damage, that can be done in several ways. One method is either to reinsulate existing laminations or to partially or completely disassemble the stack and replace them with new ones. Existing laminations can be re-coated on one side only. This decision has to be a judgment call based on experience. The coreplate material need not have any specific dielectric withstand capability, because interlaminar voltage seldom exceeds a few volts. However, two precautions are in order that may prove difficult in the service center. One is that a single layer of the material used should provide at least as much electrical resistance as the two layers present originally (one on each of the abutting lamination surfaces). Secondly, the insulation layer must not be thicker than the original, unless the number of lamination
layers recoated is fairly small. Otherwise, the core stack length can be unacceptably increased. Also, the material used for recoating obviously has to withstand motor operating temperature. Removing and restacking segmented laminations requires great care, because the segments may be of a number of different configurations that must be re-stacked in a particular order. (See "The importance of slot combination in a-c motor design" in EA June 2005.) The second method of repair has been loosening of the core clamping, then giving the core a low-viscosity VPI treatment (retightening the stack before the resin is cured). One drawback to this method is the risk of destroying that insulation if the winding fails again later and must be removed in a burnoff oven. An alternative is slipping thin mica sheets between laminations. That's of limited value for badly damaged cores, because of the difficulty of placing the mica over sizable areas, and the thickness of the material (measured in thousandths of an inch whereas coreplate coatings are much thinner). All these procedures are not only labor-intensive; they will be impossible with some machines. When large motor laminations have been effectively short-circuited together only in or around the ID or slots, particularly by machining or a rotor rub, acid etching is a common means of repair. For the bore itself, careful grinding of the surface may remove smear and adequately separate the lamination edges. This cannot, however, reach down into the teeth themselves. Several methods have been suggested, usually involving a weak solution of phosphoric acid as the etching agent. Following is the draft of one of the more elaborate procedures as proposed for an IEEE standard 20 years ago: "Make up a solution as follows: 20% by weight of 85% phosphoric acid 20% by weight of denatured alcohol 60% by weight of water Make up an applicator from dacron or wool felt fitted to a steel rod with an electrical contact and insulating handle. Soak felt with solution and apply to core by wiping the felt in the direction of the laminations. Apply variable voltage from a 120 volt a-c source between applicator and core; increase voltage until current - 2-3 amperes per square inch of applicator area. Continue application for 60-90 seconds followed by 30 seconds' cooling time to extend applicator life. Continue until clear demarcation between laminations is visible using 5% to 7% magnification. Wipe off excess acid solution with dry cloth." Chemical processes
The IEEE motor repair standard, No. 1068, describes several methods of core repair but doesn't include any such process. Etching with acid ("orthophosphoric") alone has also been done, with the caution that "care must be taken" and "only experienced personnel" should do the job. Today, of course, safe handling practices and personal protective equipment requirements go well beyond what was typical even in the 1980's. And experienced repair personnel have warned, "one out of three of these grinding or etching exercises makes the core look prettier but actually worsens the fault." For many years, the chemical process known as bonderizing has been used as surface preparation for painting steelwork exposed to outdoor or corrosive environments. Also sometimes termed phosphatizing, this process involves the chemical conversion of a steel surface into a coating to which paint bonds exceptionally well. One version of the process, known as Parkerizing after its developer, creates a rust resistant surface of a characteristically blue gray color. The typical "phosphate coating" process has been described as applied "by brushing, spraying or prolonged immersion in an acid orthophosphate solution . . . composed of four parts: detergents, metal phosphates (such as zinc or manganese phosphate), phosphoric acid, and additional agents (such as reaction catalysts or chemicals to tie up byproducts). . . . The phosphoric acid provides for minor acid etching or pickling allowing for better paint and lubrication adhesion. The metal phosphates react with the surface . . . to form the corrosionresistant phosphate coating." In a recently tested stator core repair method based on that concept, a solution containing acid with metallic compounds in solution and heated to 80°C to 100°C is applied to the similarly preheated core either by dipping (if the core is small enough) or spraying (see Figures 13-15). In a typical test on a 20 hp stator that had been through a burnoff oven, the initial core test showed a loss of 5.9 kilowatts per pound. After treatment, followed by a water rinse and dryout, the tested loss dropped to 4.1. After concern was voiced about the possible effects of conductive residue within the laminations, the core was thoroughly cleaned again, and the loss dropped further, to 3.3. Surface resistance tests on individual laminations have shown that the treatment does restore interlaminar resistance to a degree warranting continuing investigation. Significant damage to cores of the size in Figure 13, of course, will seldom be repaired. Motors of that size are typically considered throwaways unless they're of special design. Large
machines, on the other hand, will be repaired despite high cost. A 12,500 hp motor stator containing more than 34,000 individual lamination segments in a core 85 inches in ID and 80 inches long was found to have numerous unacceptably hot spots prior to rewind. The damage apparently took place when a torch was used to remove some insulation, followed by blasting with a conductive abrasive. Normally, the entire core would have required restacking. The new acid treatment process was successfully used to largely eliminate the hot spots (as shown by infrared thermography) and lower the tested core loss from 2.3 to 1.9 watts per pound, with a one-third reduction in hotspot temperatures. The process can be completed in a day, whereas disassembly and restacking could take weeks. In summary: most motor failures involving more then superficial winding damage can be suspected of having caused stator lamination damage. No rewind should be undertaken without careful checking of core condition, and economic evaluation of several possible repair methods. © 2006 Barks Publications Provided by ProQuest LLC. All Rights Reserved. Clue
