FAILURE OF MODERN GENERATORS

Clyde V. Maughan, President Maughan Generator Consultants Schenectady, NY, USA

ABSTRACT: Making generators at lower cost has had a significant negative impact on the current fleet of generators, units less than about 30 years old. A number of factors have come into play, including the "reinventing old problems", designs that push duties to higher and uncharted levels, and pressures to manufacture new machines more quickly and with less costly materials and processes. Couple these competitive market realities with reduction in number of engineers in OEM organizations and the loss of institutional knowledge as elders have retired, and it is not surprising that some machines are failing much sooner than historically expected. This paper will present examples of some of the more critical missteps for various 3600 (3000) and 1800 (1500) RPM generators and propose practical maintenance approaches for power plant engineers.

High Electrical Stress on Stator Windings Historically the electrical stress across the stator bar groundwall has increased very slowly over time. This is because electrical duty increases at about a 9th power of stress volts/mil (vpm). On the asphalt/mica windings, stress was about 45 vpm. With the introduction of polyester/mica systems in the 1950s, stress was increased to around 54 vpm. With the introduction of epoxy/mica systems in the 1960s, stress initially went into the 65 vpm range. Designs up into the 90 vpm range are now being produced. Consider that a 20% increase in vpm increases duty by1.29, or about 500%, and serious partial discharge and early wear-out problems can be expected on these windings. High Temperature Stator Winding Designs The evolution in design temperatures has been slow (and not always honest) on stators. The asphalt stator windings were called Class B (130C), but were barely Class A (105C) capability. The polyester and early epoxy windings were rated Class B, and were an honest Class B capability. Presently, usually stator windings are (mostly an honest) Class F (155C), although specifications commonly call for Class F insulation operating at Class B design temperature. The latter is a reasonable approach, since partial discharge and mechanical duties are both significantly impacted by increased temperature. Stator Bar Slot Vibration Large indirectly cooled generators tend to be made with many stator slots and tall, narrow bars. This design approach maximizes the area for transferring heat through the groundwall insulation. But it also allows the tall, narrow bar to vibrate sideways in the slot, probably driven by the core vibration. This vibration results in vibration sparking, a fast wearing destructive phenomenon.1,2 Photos 1 & 2.

PREMATURE FAILURES OF MODERN GENERATORS Historically utility-size generators were made with considerable margin in load capability. With the simultaneously advent of powerful computers and greatly increased competitive conditions between Original Equipment Manufacturers (OEMs), this margin was largely removed. These new designs increased mechanical, electrical and thermal duties to a point where there is little margin left, and occasionally generators are produced with designs that exceed specification parameters. The result is a fleet of generators that increasingly require major maintenance long before the historic target of 30 years for stators and 20 years for fields. The following paragraphs will cite some examples.

Photo 1 Side of a Bar Deeply Damaged by Vibration Sparking

Photo 4 Several inches of copper burned away on a stator bar

Dry Endwinding Support Ties The endwinding support systems on most generators rely on bonding of ties to the support structure and stator bars. Some OEMs have used a wet tie to make this tie as strong as possible. (Wet meaning that the glass for making the tie is drawn through a high bonding strength resin just before the tie is made.) For cost and simplification reasons, a changeover was made to dry ties. (Dry meaning that the glass is pre-impregnated with a partially cure resin that is dried to allow for ease of application.) These dry ties have not performed well and have led to numerous repairs involving removing of the dry ties and replacing with wet ties. Photos 5 & 6.

Photo 2 Side packing almost completely eaten away by vibration sparking

It has been supposed by designers that side filler would not allow side vibration, but since packing cannot fill the space fully, vibration is allowed. Corrective action probably will require rewind to remove the fatally degraded bar and allow installation of side pressure springs. Stator Endwinding Vibration Endwinding vibration has been an ongoing problem on large generators. Problems have accelerated as engineers have designed for higher vibration driving forces and simplified support systems.3 Eliminating series blocking on an endwinding resulted in the broken bar shown in Photo 3.

Photo 5 Dust generation from failed dry tie

Photo 3 Broken series connection

Local endwinding resonance resulted in the failure shown in Photo 4.

In at least one generator, an arc occurred at a failing flexible lead. The resulting arc destroyed the insulating

properties of the atmosphere and led to massive arc damage at the bared copper. Replacement of the stator was required to repair the damage. Series and Phase Connection Insulation The connections are difficult to insulate with mica tapes. A much faster and lower cost method, the use of a non-mica potting compound, is common. This is perfectly adequate at the low voltage difference of the series connections.4 But not necessarily so at phase-to-phase locations where partial discharge can occur. Photos 7 & 8 are at one such line-to-line phase break.

Water-cooled winding connections can be fully insulated. But on modern generators, designers have occasionally omitted insulation as a cost reduction. The result of this small cost reduction on an 800 MW generator has been a failure similar to Photo 9. Vacuum Pressure Impregnation. Vacuum pressure impregnation (VPI) of individual stator bars with asphalt began some 97 years ago. Stator bar VPI with thermoset resins began about 1949 and has been successfully used to make high quality stator windings since that date. It was soon recognized that Global Vacuum Pressure Impregnation (GVPI) would be a fast and inexpensive method of making a stator. (With GVPI the entire stator wound with dry coil insulation is placed in a very large VPI tank. The entire stator winding is impregnating in a single operation.) Because of these cost benefits, GVPI has long been used on motors and smaller generators. Beginning about 1975, GVPI became popular with some OEMs for making large generators, up to the range above 400 MVA. These windings have been subject to some early maintenance issues, including: Slot partial discharge and vibration sparking, which seem to be related to the slip plane between bars and core. Endwinding vibration, which in some designs may be related to a system that does not allow for the inevitable axial expansion of the bars out of the ends of the slots. Difficulties in performing a rewind, which result from the VPI process bonding the stator bars into the slots (irrespective of the slip plane). It is difficult-toimpossible to remove the winding. Where rewind is performed, the process is slow and the core is vulnerable to damage. On a GVPI generator, stator replacement may be preferable to attempting rewind. Notwithstanding these limitations, because of the major advantages to GVPI, this process will continue to be popular for large generators. And it is certain that methods will be found to address the inherent problems associated with GVPI. Step-down at End of Core. Because of high design uprate, it is necessary to have a much greater step-down at the end of the core. Because the top bar now is above the core iron, there is a significant electromagnetic sideforce driving side vibration on the top bar. This seems to be resulting in vibration sparking and other concerns on the winding. Photo 10.

Photo 7 PD at a non-failed phase connection location

Photo 8 Failed connection after removing insulation

Designers now typically use potting compound on the series connections, but tape the phase connections with mica tapes. On direct-gas-cooled stator windings it is impossible to fully insulate the connections access for gas flow must be permitted. Numerous massive winding failures have resulted. Photo 9.

Photo 9 Ring-of-fire failure of a stator endwinding

temperature. These already poor mechanical properties begin to fall off rapidly as the temperature of the copper increases above about 130C. As a result, problems on field windings have tended to remain rather constant over the years, e.g., broken turns, grounds, turn shorts, coil shorts, coil migration, distortion of turns and coils, broken pole-to-pole and turn-to-turn connections, slot and turn insulation migration. Forging cracks have occurred, and there has been an occasional catastrophic forging failure. But these tend to be unrelated to cost reductions. Direct cooling has been the greatest uprate tool on fields, and direct cooling does bring problems, primarily due to cleanliness issues in contaminated atmospheres. However, this system has been used for 50 years, and the problems have remained about the same for 50 years.

Photo 10 Dust generation probably resulting from vibration due to side forces on top bar

Tape Migration on Stator Bars. This was a highly destructive deterioration mechanism on asphalt windings in large (>40 MW) pre-1960 generators.5 Photo 11.

PERSONNEL ISSUES RELATING TO PREMATURE GENERATOR FAILURES Inherent Personnel Limitations Undoubtedly cost pressure is the primary cause of increased failure rates of modern generators. Not only have these pressures resulted in less design margin in the generator, but have also seriously reduced the strength of the OEM generator design, manufacturing and service organizations. A further negative factor relates to the inability to pass on to the younger generation the knowledge of the previous generation of design and service engineers. Because much of the knowledge of how to build a generator is in a real sense an art, this has repeatedly led to reinventing old problems. A case in point is that of dry ties discussed above. Or failure to correctly make up a bolted electrical joint, leading to massive stator winding failure. Under existing industry conditions, it has simply not been possible to maintain the skill levels and quantity of talent to adequately perform the required design and assembly tasks. There are secondary contributors to the degradation of inherent talent. Three come to mind: 1) power generation is no longer a glamor business for attracting the needed top talent demanded for production of the complex and not well understood generator, 2) the present generation of design engineers is spread so thin it is difficult for any individual engineer to become highly skilled in any one discipline, and 3) the advent of computer-aided design had reduced the feel for the design that came from designing with a hand calculator and a 10 log-log, deci-trig slide rule. Two incidents may illustrate the fundamental weakness of OEM organizations:

Photo 11 Typical tape separation (girth crack) on a pre-1960 armature bar

This migration was eliminated by use of thermoset windings. But the problem has recurred on post-1970 small (<20 MW) coil-wound generators that were made with a low-grade asphalt material. Photo 12.

Photo 12 Remaining groundwall measured 20% of original thickness

Rotating Field Problems. Design margins in fields have not been as impacted as greatly as those on stators, largely because there never has been a great amount of margin on fields. But also, since copper mechanical properties are so marginal, there has not been much opportunity for cost reduction and increase in duties on fields. Design temperature on fields has been fortunately slow to evolve upward. High temperature insulation materials are available. But designers recognize that copper has inherently low copper mechanical properties even at room

A field had been forced out of service by a thermal vector of unbalance resulting from axial migration of 3 coils in the slot of the field body. The key OEM field design engineer is at the white-board explaining to the owner why this migration was causing the vector. He also explained that the axial blocks under the retaining ring prevented this migration. Of course, it is the peripheral blocking that prevents migration, and this engineer could not be made to understand this obvious design feature. End-of-slot vibration occurs on some generators. This vibration results from incorrectly setting the endwinding support system such that the bars are held off the bottom of the slot. The high electromagnetic downward force on the bars results in the bars destructively vibrating at the ends of the slots. The key OEM engineer was absolutely unable to visualize this condition and continued to believe that the problem resulted from oil contamination of the winding.

Another fundamental personnel problem results from the OEM, and independent, generator specialists being unwilling to say: I dont know. These generator specialists (from OEMs or independent service providers) are trained to perform inspections, tests and maintenance on generators. Few have the technical and experience basis on which to perform the extremely challenging failure root cause analysis. Unfortunately, few specialists seem willing to say, I dont know, and all recognize that it normally will be very difficult to obtain assistance of a qualified, capable engineer from the OEM engineer organization. (See comment above.) The result is often incorrect diagnosis of major failures on large power producers. This can be exceptionally costly to the owner. Secondary Impact of Personnel Limitations Inadequate Failure Root Cause Diagnosis

somewhat intuitively obvious to a knowledgeable, intelligent person, there is little about a generator that is obvious to even the best of intuition. b) Generators can fail in many failure modes. Plant personnel will have heard an OEM engineer say: We havent had this kind of failure before. This may be rightfully greeted with some skeptical. But in fact because there are so many ways a generator can fail, this comment may be quite true and accurate. (After 62 years of generator experience, the writer is currently involved in such a situation.) c) Typically no one in the power plant knows much about the function of the generator. In a power plant there will be several operating and maintenance personnel that quite clearly understand the turbine. These personnel often can greatly assist in determining why a component has failed. This is unlikely to be true of plant personnel with respect to the generator. d) Finally, even with the best generator monitoring instrumentation, often the plant records will shed little light on the failure root cause. Mis-diagnosis of the root cause of generator failures can be exceptionally costly in that corrective actions may not address the actual cause leaving the generator vulnerable to repeat of the failure. The direct repair cost of such a failure can range from a few hundred-thousand dollars to many millions of dollars. In addition, loss-of-generation costs can exceed $200 million on a major failure of a large generator. Five real-life examples 1. Ring-of-fire stator winding failure on a 900 MW watercooled winding was incorrectly charged to the piping arrangement. Had the correct error not been identified by further consultation, repeat of the massive stator winding failure would have been inevitable. Photo 9. 2. Stator winding failure on a 200 MW stator incorrectly charged to lightning. Corrective action taken was based on this mis-diagnostics. The unit failed again several months later at the same location for the same cause partial discharge. Photos 7 & 8. 3. On a 250 MW generator, contamination of a core endpackage was judged to be from minor stator bar vibration and no corrective action was taken. The root problem was actually local core looseness, and the winding failed to ground a few weeks after return to service. Photo 13.

Historically diagnosis of new and complex generator failures was done by original equipment manufacturer (OEM) factory engineers. This was the case because accurate diagnosis can be significantly assisted by a technical education background and by generator engineering design experience. But with the contractions in number of OEM expert engineers, there are now available few OEM engineers with good diagnostics capability. Because personnel of this caliber are vital to other OEM efforts, obtaining their services at a power plant may be a challenge. Thus it has become increasingly difficult to obtain personnel that are capable of performing diagnostics. But root-cause diagnosis of generator failures is a particularly challenging task. Specifically: a) Generators are complicated mechanically, and more particularly, the theory of the function of a generator is complex. While design and operation of most mechanical components of a power plant are

Photo 13 Knife check of locally loose core

4. Core discoloration over-looked at location of stator winding failure. This 80 MW stator was rewound without correcting the core-iron condition. New replacement winding failed in service shortly after return to service. Photo 14.

future. The challenges will continue, and it is hoped that the above information will be of some assistance in handling these challenges. References: 1. 2010 IEEE International Symposium on Electrical Insulation, San Diego, California. Insulation Degradation in Generator Stator Bars due to Spark Erosion and Partial Discharge Damage, William G. Moore and Aleksandr Khazanof, 2. IEEE Electrical Insulation Magazine, September/ October 2008 Vol. 24, No. 5.G. C. Stone, Iris Power, et al, Impact of Slot Discharges and Vibration Sparking on Stator Winding Life in Large Generators. 3. EPRI Winter Technical Workshop and TGUG Meeting, January, 2005. St. Petersburg, Florida, USA. John Demcko & John Velotta, Arizona Public Service Company, Phoenix, Arizona, USA. Generator End Turn Vibration Monitoring. A Case Study. 4. Electrical Insulation for Rotating Machines, G.C. Stone et al, Wiley-IEEE Press, 2004. 5. EPRI International Workshop on Maintaining the integrity of Generator Cores, San Antonio, Texas, 2001. Riverbend 4 & 5 Core and Coil Failure, Myron Horton, Gary Sloviski, and Bill Moore.

Photo 14 Local core overheating

5. Asphalt stator winding incorrectly diagnosed as having destructive migration of the groundwall. Photo 11. An inexperienced OEM engineer, and an inexperienced independent consultant, each recommended immediate stator rewind to avoid catastrophic service failure. Their recommendations were voided by a more experienced engineer, and generator continues to run safely many years afterward.

CONCLUDING OBSERVATIONS There are no easy answers to the dilemma of marginal equipment and inexperienced service personnel. Generally a deficient design does not leave margin for modification to a lower duty design. The situation on service personnel limitations is not subject to ready correction. Unfortunately, the difficulties and challenges will remain until such time as the industry reduces the focus on first cost and addresses the long-term costs of power generation equipment. Until that utopian condition arrives, the challenge to power generation producers can perhaps be summarized as follows: 1. Monitor the generator with state-of-the-art equipment. 2. Keep the monitoring equipment in good operating condition. 3. Be certain to record and retain all monitoring equipment information. 4. When failure occurs, assure that the failure investigation is thorough, and that the diagnostics information being provided from on-site expert personnel is plausible and reasonable. 5. If it is not, obtain additional technical support, and do this early, before critical information and valuable time are lost. It seems clear that the problems facing the power generation industry will not be cleared up in the near