Ryan LeClair, generator product-line manager, Gas Power Services, opened the GE Day program for the Generator Users Group (GUG) with an overview of the topics to be discussed, including: negative-sequence-current impacts and causes, fleet updates and new observations, 7FH2/B family and 2020 stator update, amortisseur spring axial migration cases and retention solutions, HiPot testing, and third-harmonic stator ground. Highlights of the presentations made November 19 during the second week of the 2020 virtual meeting follow. To dig deeper, access the presentations on GE’s MyDashboard.

“Negative-Sequence-Current Impacts and Causes.” Janusz Bialik, principal engineer, Generator Product Service, and Ben Mancuso began by outlining their presentation: Current induction theory, overview of damper systems, case studies, and recommendations.

Negative sequence currents are a common and destructive phenomenon in generators—generally preventable by good design and operation, and the use of monitoring equipment. They can be caused by unbalanced three-phase currents, unbalanced loads, system faults, open phases, asynchronous operation, etc.

Recall that the phase currents and voltages in a three-phase power system can be represented by three single-phase components: Positive-, negative-, and zero-sequence. The first has the same rotation as the power system and represents balanced load. The second has a rotation opposite that of the power system; the zero-sequence component represents an unbalance that causes current flow in the neutral.

The negative-sequence current component circulating in the stator windings creates a magnetic flux in the air gap of the machine. This flux rotates at synchronous speed, but because its direction is opposite to the normal flux, eddy currents are induced in the rotor body. They tend to flow mainly in the outer regions of the rotor, because of the “skin effect,” which was explained by Bialik.

If large enough, the induced currents will spark and arc between wedges, wedges and forging, wedges and retaining rings, forging and retaining rings, and any component on the periphery of the rotor. Such sparking/arching can cause hardening of the metal in critical areas, followed by cracking.

Thus, negative-sequence currents have the potential to cause rapid heating and significant damage to the generator, dictating the need to protect against them. Loss of mechanical integrity or insulation failure can occur quickly.

Generator designers include many features to mitigate the effects of negative-sequence currents, Bialik told attendees. Examples include assuring good electrical contact between rotor structures to prevent arcing, a damper system (a/k/a amortisseur windings) in the rotor slots to form low-resistance paths across the rotor surface, aluminum slot wedges, etc. Use of a damper bar in the pole zone of the machine and a low-resistance endwinding design are illustrated in the presentation available on the Power Users website. [link]

“Generator Fleet Updates and New Observations,” Ross Sacharow, generator product service manager, began with an overview of what he considers one of the OEM’s most important—possibly the most important—maintenance document, GEK103566 Rev M. It includes a new section dedicated to brushless-exciter maintenance. This just-released guide is a complete update of Rev L, combining input from both the GE and Alstom fleets.

A particularly valuable slide for users, the fourth in the presentation, details the terminology changes important to owner/operators and defines first, borescope, and rotor-in robotic inspections.  Access the presentation on the Power Users website. [link]

Sacharow told attendees that where inspections with robots are possible (air gap large enough to accommodate the robot), a rotor generally should be removed only for repairs and upgrades. He cited the following among the maintenance/repair activities that would require rotor removal:

• • • Rotor tooth and wedge inspection and repair to remove hardened material.
    • Rings-off cleaning up to field rewind.
    • Partial or full stator re-wedge.
    • Core or wedge repair.
    • Partial or full stator rewind.
    • Retaining-rings-off inspection.

Another slide summarizes inspections and maintenance intervals as specified in Rev M alongside those published in Rev L with updates highlighted—a handy guide for maintenance managers.

GE polled the group to learn how users view the ability of robotic rotor-in versus rotor-out inspections to find a problem—assuming no known components to repair. Here’s what they said:

• • • Very comparable, 13%.
    • Similar, 25%.
    • If it were up to me, I would always pull a rotor for a major inspection, 42%.
    • If it were up to me, I would always use a robot for a major inspection, 20%.

Recent findings regarding collector flashovers was Sacharow’s next topic. He said reducing selectivity—the unwillingness of brushes arranged in parallel to share current equally—was critical to reducing such incidents. Much has been published on the subject, yet flashovers still occur—typically a dozen incidents worldwide annually. [link to https://www.ccj-online/maughan/brush-collectors/] For those requiring a refresher, the presentation has a slide dedicated to collector-system maintenance activities (daily, weekly, monthly, during an outage). This might be a good topic for a lunch-and-learn in the plant break room.

Next topic: Recent 7FH2 core-damage findings. At least six examples of turbine-end core damage at the step iron have been reported—five in the last year. All were found on generators made by a third-party. In each case, there was evidence of core ID looseness at the teeth. Risk is not clearly tied directly to operating hours or starts. Preliminary findings point to manufacturing practices as the cause of the looseness.

With the root-cause investigation ongoing, the following recommendation was made for units in this fleet not planning a generator outage in 2021: Perform a borescope inspection—as early as possible in the year—focusing on visual evidence of turbine-end step-iron core damage or greasing/red dusting.

For units planning a generator outage in 2021, suggestion is to follow the standard inspection procedures presented in GEK103566 and use the outage to thoroughly inspect for evidence of core looseness or lamination damage. The recommended inspections (a list is included in the presentation) require removal of end shields, gas baffles, and at least one cooler. Questions? Contact your GE service rep.

“7FH2/B Family, 2020 Stator Update.” Eric Buskirk, generator systems integration leader, responsible for guiding new unit and technology development, covered the rotor/stator configurations and manufacturing variables associated with this fleet, plus stator features, generator efficiency, fleet-wide issues and updates, and cyclic duty.

Presentation began with clarification of product nomenclature important to owner/operators. 7FH2 generators, made from about 1994 to 2015, dominate the fleet with about 850 units in service—the overwhelming majority coupled to 7F gas turbines. This product, renamed H33, has been phased out. Replacement is the 7FH2B, recently renamed H35, which has been manufactured since 2004. There are 75+ units in this fleet segment.

Users should be aware that during the gas-turbine bubble (1999 to 2005), GE expanded its generator manufacturing capability through partnerships with Doosan, Alstom, Melco, Toshiba, and Hitachi. While all insulation systems and generator processes were qualified by GE, there are minor manufacturing differences among the machines produced under license; some parts are interchangeable, some not—stator cores, for example. Valuable detail is provided in the presentation available on MyDashboard.

Buskirk mentioned the availability of exchange fields for the H33 and H35, which could yield a small capacity uprate, noting that the stator may be limiting and auxiliaries may have to be upgraded in order to capture the full-capacity uprate entitlement. Generator details (leads up versus leads down, coolers and their arrangement, bushing design, connection-ring configuration, loop blocking, etc) are explained in a series of slides easily understandable by O&M personnel.

Generators generally are very efficient, the speaker reminded, but there are opportunities for small, but meaningful, performance improvements under certain conditions. Example: Cooling may be improved by raising hydrogen pressure, but at the expense of increased windage/fan losses. If a machine is running cold or not operating near nameplate, perhaps a reduction in gas pressure would be beneficial. For an H35, going from 45 to 30 psig saves about 75 kW. Raising hydrogen purity (by reducing seal-oil flow and making other adjustments, for example) can save up to another 100 kW or so. Several more possible improvements were explained as well.

“Generator Field Amortisseur Spring Axial Migration Cases and Retention Solutions,” Ben Mancuso, generator design engineering technical leader. The purpose of the amortisseur winding is to divert the negative-sequence current (see first presentation summary above) from flowing in the rotor forging and causing arcing and/or overheating damage during normal operation. Be aware that some abnormal events can exceed amortisseur capability.

Mancuso provided valuable graphics and summary statements to explain the various slot amortisseur windings (Gen 1, 2, and 3, plus non-static start) found in the fleet and how they work. Plus, he addressed excessive axial misalignment—when a wedge vent hole exceeds 30% or more blockage—and its causes. Should wedge hole blockage exceed 50%—not typical—it’s possible to cause a hot spot in the winding, possibly leading to an electrical fault.

Finally, the speaker presented solutions to address component migration—including an amortisseur/spring axial retention modification package available from the OEM to lock the various components together.

“Over-potential Insulation Dielectric Testing (HiPot),” Michael Villani, product and factory support, Generator Services. Periodic maintenance over-potential testing of generator stator windings is a generally accepted industry practice. The speaker said that when this test is performed on healthy insulation and according to the OEM’s instructions, the insulation will not be damaged. Not everyone in the industry would agree with this. However, if the insulation’s integrity has been degraded prior to the test—such as from stator-bar abrasion, girth cracks, etc—the test will accentuate that damage.

Later, Villani noted that the final hi-pot test during generator manufacture is considered the most reliable factory test to determine the quality of the insulation system at the time of shipment. Thus, he concludes, continued periodic hi-pot testing in the field is the most reliable test to determine the suitability of an insulation system to perform its intended functions reliably. The customer ultimately decides what voltage it is most comfortable with for conducting the hi-pot.

Villani also reviewed the pros and cons of ac and dc hi-pot testing, noting that all new stator bars and stator windings are tested with ac hi-pot. With the ac test set large and requiring trucking and cranes to transport and move around onsite, and a significant power source, its cost is higher than for the dc test, which is selected in most instances.

Final topic in Villani’s presentation: Should hi-pot testing be closed or open? GE generally recommends hi-pot testing be performed with the unit degassed and opened up, so both ends of the generator are visible to the operator/second watch person and others observing the test. The benefit of this approach is that if a failure occurs it may be possible to see its location, thereby helping to quickly identify the failed winding.

“Third-Harmonic Stator Ground,” Dhruv Bhatnagar, generator application engineering. This was the second presentation on the topic at GUG2020, the first delivered by an owner/operator during the Week One generator session. Content of both presentations is similar and those with interest in the subject matter should review both—the user’s on the Power Users website, and Bhatnagar’s, which provides GE’s recommendation for stator-ground protection, on MyDashboard. An important point: GE recommends simple and low-cost testing to ensure that the 27TN trip settings are properly configured. If the are not properly configured, significant generator damage can occur, as the examples provided in the presentation revealed.

The annual conference of the Generator Users Group is being conducted online for the first time this year. The four-week event, exclusive to owner/operators, began November 12 with a program focused on user experiences and vendor presentations. The latter were limited to 30 minutes each and conducted in two half-hour sessions. Each vendor was assigned a “breakout room” and users were connected to their presentations of choice. While attendees could not participate in more than two vendor presentations during the live Week One program, all presentations—both user and vendor—are available now on the Power Users website, mostly in video format.

Highlights of the Week One user and vendor presentations, developed by the editors, are below.

Week Two (November 17-19) was GE Week, characterized by meaningful technical presentations, Q&A, and open discussion on topics of interest to the user community. Those presentations can be accessed through the OEM’s MyDashboard website.

Week Three’s program begins on Thursday, December, 3 at 9 a.m. Eastern with the first three hours allocated for private meetings with vendors; schedule these online just as you would a doctor’s appointment. Presentations by users and independent consultants are next, from noon to 2 p.m. followed by these two hour-long live vendor presentations with Q&A:

• • National Electric Coil, 2 p.m., “Corona in High-Voltage Stator Coils: Theory, Causes, Repair, and Laboratory Prognosis” and “High-Voltage Generator Stator Ground Insulation Repairs.”
  • MD&A, 3 p.m., “Generator Findings and Case Studies” and “Preparing Generator Rotors for Cyclic Duty.”
  • A virtual vendor fair, complete with video chat rooms, runs from 4 p.m. until the end of the day’s program at 5.

GUG2020 concludes with the Week Four program on December 10. Program format is the same as that for December 3. The featured vendor presentations on the last day of the conference are these:

• • AGT Services, 2 p.m., “7F Simple- and Combined-Cycle Generator Field Repairs.”
  • Siemens Energy, 3 p.m. “Case Studies: Cycling Impacts on Generators,” “Third-Harmonic Stator Ground Protection,” and “Hipot Testing approach.”

GUG2020, Week One user presentations

“Subharmonic Stator Ground-Fault Protection” is an important presentation for generator owner/operators because the failure to protect adequately against ground faults could lead to a hugely expensive outage in terms of equipment damage and prolonged loss of generation. Unfortunately, to fully understand the protection schemes available to prevent such damage it helps to be an electrical engineer with power experience—a rare species in an industry dominated by mechanical engineers.

Perhaps no one knows this better than Clyde V Maughan, who has degrees in both mechanical and electrical engineering, and was the muscle behind the founding of the Generator Users Group in 2015. He “blew the whistle” on the shortcomings of IEEE standards in adequately protecting against grounding issues in stator windings back in mid-2013 with his article in CCJ, “Generator Protection: IEEE standards may not sufficiently address grounding issues in rotor, stator windings.”

The GUG2020 speaker reviewed some material covered in the Maughan article and provided details on a ground-fault experience that involved two unit trips several months apart. The second fault occurred because analysis of the first trip proved incomplete. This was a real “whodunit” in a sense because questions remain; however, the generating unit has been back in service for more than a year with no issues reported.

The presenter reminded attendees that large generators have high-impendence grounding (neutral) connections to limit ground-fault current (read “damage”). Other generator ground-fault protection methodologies require the unit to be online. The two methods typically used are the following:

• • 59N—Generator Neutral Over-voltage Protection, which detects faults in the top 95% (approximately) of the generator up to the generator transformer.
  • 27TN—Third-Harmonic Neutral Under-voltage Protection, which detects faults on the neutral side of the generator.

Together 59N and 27TN can provide full protection for the generator but many units still are protected only by the 59N relay—the 95% solution—and are at risk. Consider, too, that some generator designs may not produce sufficient third-harmonic voltages to allow reliable ground-fault protection schemes based on third-harmonic signals.

Given these concerns, voltage-injection relay systems (64S) have gained favor—especially where there are concerns regarding stator-winding condition and/or when the unit is deemed critical by the owner. A big advantage of the 64S is that it provides its own excitation source and assures the entire generator ground-fault protection when it is shutdown (at standstill or on turning gear), during startup, and at-speed offline or online.

Concerns regarding 64S are cost, more complicated analysis of issues, and more equipment to go wrong.

“Remote Technical Support during Generator Maintenance,” the second presentation at the conference, was insightful given the dearth of generator engineering talent in the electric power industry today (see previous article), and travel and other restrictions that have become the norm during the global pandemic. It was developed by Clyde V Maughan and would have been the 94-year-old’s first formal virtual presentation had a schedule conflict not interfered. Jim Timperley, well respected by generator users worldwide, made the presentation.

Timperley began by illustrating the need for, and drivers of, remote technical support, complete with examples. Next, he covered the advantages of working remotely—saves a lot of money and time. And the disadvantages: It’s difficult to see and hear about everything you might want to know without the close personal connections typically developed during a site visit. Regarding the last point, Timperley cemented Maughan’s conclusion that this will be less of a concern in the future because of the rapid advancements in digital technologies and communications.

To learn more about what Maughan has to say on remote support, read his article in CCJ No. 64, p 2.

But wait, there’s more. With retirements, cutbacks in plant staffing, and the absence of traditional face-to-face conferences, Maughan thought it necessary to have a consultant register as a means for introducing generator-owner/operator personnel to candidate experts capable of fulfilling their needs.

So he worked with General Manager Scott Schwieger to develop the Generator Consultant Skill Register, posted on CCJ’s website, where you can search for an expert with the specific skills required for your assignment. If you’re a generator expert who wants to be included in the register, present your qualifications for vetting at www.ccj-online.com/generator-expert-skill-register.

“When a New Stator Requires a Rewind: Vendor Oversight Lessons Learned” is a case- history that every user should read through. It tracks oversights/errors compounded over a period of six months and offers lessons for all plant personnel that don’t just apply to generators.

The “adventure” begins during initial startup testing: Generator vibration increased and the turbine tripped. Generator fans had been installed on incorrect ends of the machine and a blade liberated. There was no alert from the M&D center because it did not monitor units in commissioning. Damage to the machine was minimal, essentially limited to deposits of foreign material sprinkled throughout the generator. The unit was cleaned and the unit restarted relatively quickly after appropriate inspection and testing.

Likely, those involved breathed a sigh of relief at that point thinking they had dodged a bullet. Well. . . . About a month after the first blade failure there was another, but this one was more severe, with over 500 damage locations on the endwinding. An engineering assessment based on bar voltage and damage location and depth revealed 14 bars at high risk, 10 medium risk, and three low risk.

The root-cause analysis of this failure revealed the blades that had been installed incorrectly on the turbine end of the generator before the first failure had been reused on the collector end. New blades were installed only on the turbine end after the first event. Engineers found that because the reused blades had operated backwards, they were prone to cracking from high-cycle fatigue. No NDE was performed on the blades prior to their reuse.

The plant owner decided on bar repair to fix the damage done. Following lab testing of the proposed repair method, it was implemented on the affected generator. A follow-up ac hipot test failed on two phases, but neither failure occurred at a repair. Lab analysis determined the failures were caused by subsurface cracks attributed to debris impact. No visible surface damage was in evidence at these locations.

Next, a decision was made to proceed only with the replacement of the two failed top bars. But during this activity, six bottom bars were damaged by metal wedges. A full rewind was ordered.

Now four months into the project, during assembly of the stator bars, misalignment between the top and bottom bars was discovered. The bars originally were manufactured on the wrong bar form for the stator frame. When the top-bar replacements were planned, as-built dimensions were used instead of the design values.

You can only imagine what went on from that point until the generator was cleared for full-load operation about two months later. Get the details in the presentation, available to users only on the Power Users website.

Lessons learned were plentiful on this project, including these:

• • Written communication between owner engineering and vendor field services is necessary for important recommendations and decisions. Too much can be lost or forgotten in verbal communication. Had this been done for the fan-blade replacement activity after the first blade failure, the second failure might not have happened.
  • Owner oversight of critical activities is very important. It’s highly unlikely that an experienced owner’s eyes would have missed use of the destructive wedges during top-bar removal.
  • Carefully vet the vendor’s human performance program regarding adherence to procedures, validation of assumptions, staff communication, and training. Don’t expect that the vendor’s human performance program is as rigorous as yours. Certain things were assumed during the manufacturing phase of this project that should have been verified before moving forward.
  • Request shop quality exceptions. Recall that incorrectly manufactured bars were discovered during the rewind. There was no notification from the vendor during original manufacturing of the generator. The owner was told later such notifications had to be requested. Important to note here that the owner was not “allowed” to have a witness monitoring critical shop operations.
  • Document and maintain vendor quality performance records. Don’t expect the vendor to do this for you.
  • Be sure you have the “right” (read, most capable) people in the field to monitor contractor activities. It’s important that your team relentlessly follow up on QA/QC activities and collect and maintain meaningful data. Dotting eyes and crossing tees is vital to project success.

“Effects of Flexible Operation on Generators—Stator Winding.” Bill Moore, PE, EPRI’s technical executive for generators, began his presentation by reminding owner/operators that while the effects of stop/start operation on rotors are discussed relatively frequently, the impacts of flexible operation on stator windings have not been. He referred attendees to EPRI Report 1008351, publicly available at www.epri.com, which says the forced-outage rate of generators increased by 246% because of speed (stop/start) cycling.

In simple terms, Moore told the group that as a generator changes load its temperature changes. Given that endwinding materials (copper, blocking, insulation) have different coefficients of expansion, epoxy bonds break, blocks move and shift, dusting and greasing occurs and insulation can crack. Endwindings loosen as well.

The core of Moore’s presentation was an analysis by Siemens Energy which provides a calculated approach to accounting for stator-winding degradation in various types of generators for various load-cycling scenarios. Get the details in the presentation available to users on the Power Users website.

Moore shared the following conclusions/recommendations:

• • Repeated major load swings cause a significantly higher degradation of the stator winding than smaller load swings.
  • Areas of high degradation typically are inboard, towards the core, rather than outboard at the series connections.
  • A typical indirect air-cooled generator is likely to suffer a higher rate of degradation from load cycling than a GVPI air-cooled generator.
  • Temperature control could be used to reduce temperature change during load swings and can provide the opportunity to reduce degradation.

GUG2020, Week One vendor breakout presentations

Nel Hydrogen

Advanced Onsite Hydrogen Generation Solutions for Power Plants

Dave Wolf told attendees that onsite hydrogen production is a safer, more economical alternative to delivered hydrogen for generator cooling. He explained, with an assist from system drawings, how ultra-pure, pressurized, dry hydrogen gas is produced onsite from electricity and water using the company’s Proton Exchange Membrane electrolyser. The compact system is easy to install and available in capacities to provide scavenge hydrogen for combined-cycle plants of any size.

Wolf went on to say that the required hydrogen storage capacity for delivered gas typically ranges from 60k to 100k scf at combined cycles maintaining inventory for both scavenge and regas use. Plants opting for a Nel Hydrogen system can either produce and store regas or size the system for scavenge gas only and have regas delivered as necessary.

BPhase Inc

Monitoring the Generator with a Fiber-Based Thermal/Strain Monitoring System

Derek Hooper walked attendees through the operating principles of a fiberoptic monitoring system for providing online analysis of core temperature and coil motion in electric generators. The system, which provides a way to assess coil tightness, is a work in progress with early results described by the speaker showing promise.

Sensors, installed before wedging, provide wavelength amplitude measurements which are used to determine the temperature and motion of components. Measurements were taken in several slots during starts and full-load operation at both the turbine and exciter ends of the unit. One of the graphs presented indicated that coil displacement at the end of the core was, perhaps, as much as four times that about 30 in. from the end of the core.

Brush Group

Design and Manufacture of Non-Sparking 13.8-kV 2-pole Generators for Operation in Hazardous Areas

Roy Beardshaw told owner/operators Brush has supplied about three-dozen (in round numbers) of its DAX 2-pole generators for operation in areas where ignitable concentrations of flammable gases, vapors, or liquids are of concern but not likely to exist under normal operating conditions. Most installations are in Brazil, Nigeria, North Sea, and Asia.

The most popular drivers have been the LM2500 and SGT A35 RB; the generator model supplied for these prime movers typically is the BDAX 71-193 ERH with brushless excitation and rated for 25 to 40 MW depending on site conditions. End-frame stator design allows for quick installation and ease of access for maintenance.

Beardshaw reviewed design requirements for the machine and its safety features. The latter includes the method used to control and monitor the release of purge air into the machine prior to startup and the continual monitoring of enclosure pressure to ensure it remains ambient. If pressure drops, risking the ingress of a hazardous atmosphere, an alarm will sound and/or the machine will be shut down.

Details on testing and “explosion-proof” certification are provided in the presentation available on the Power Users website.

Electric Machinery

CO2 Cleaning of High-Voltage Windings—a Cautionary Tale

Sean Orchuk, Electric Machinery’s field service manager, reminded GUG attendees that the use of CO2 pellets as a generator cleaning tool is well-documented, but long-term insulation loss must be weighed against the potential benefits of dry-ice blasting.

Generators are expected to function for decades with minimal intervention, he said, and between maintenance cycles the stator windings may be exposed to oil and other sticky substances. There is a strong motivation to remove surface films because they attract contaminants, thereby reducing insulation resistance, increasing the propagation of corona, and ultimately causing winding failure.

CO2 pellets, Orchuk continued, offer the promise of a cleaning medium without the challenges associated with chemical cleaning and are proven to remove dust, carbon, oil, etc, from windings and other irregular surfaces. The high velocities used in CO2 cleaning processes create a risk of damage to the insulation systems of generator windings. When improperly used or misapplied, dry-ice particles can damage brittle mica and other materials used in winding tapes—an unintended consequence of the maintenance process.

In addition, the cleaning medium is limited by the line of sight from the nozzle, and significant areas of the generator winding may not receive direct benefit from this method. Case studies provided in the PowerPoint, available on the Power Users website, review cleaning methods for typical insulation systems, as well as typical typical contaminants and the range of velocities recommended for their removal with minimal or no insulation damage.

Electrical Builders Inc

Anti-Condensation Measures for Metal-Enclosed Bus—What You Need to Know

Steve Powell, EBI’s subject matter expert on bus duct, addressed the concerns of owner/operators regarding condensation in metal-enclosed bus and suggested methods for its control. His slides, available on the Power Users website, provide insights on the following topics:

• • Causes of condensation in busduct—design flaws (such as no drains in outdoor installations), poor maintenance, faulty heaters, poor installation.
  • Types of bus designs associated with condensation issues—outdoor installations without pressurized hot air for condensation control.
  • When to be concerned about condensation.
  • Solutions for controlling condensation, including the purpose of each and the type of bus where used.
  • Maintenance concerns.

If a backgrounder on isophase bus is in order before digging into Powell’s presentation, you can find one on CCJ’s website.

Nord-Lock Inc

Exciter Coupling Bolts and Solutions for Vibration-Induced Bolt Loosening

Julie Pereyra introduced owner/operators to Nord-Lock’s wedge-locking washers that prevent bolts from loosening because of vibration. They are said to reduce maintenance costs and increase worker safety. Steve Busalacchi followed with presentation on the company’s EzFit expansion bolts, which promote safe and easy making and breaking of turbine couplings as summarized in the STUG2020 segment of the Power Users report.

Get the details on both products in the presentation posted on the Power Users website.

Presentations made by National Electric Coil, MD&A, AGT Services, and Siemens Energy to owner/operators participating in Weeks Three and Four of the virtual GUG2020 conference are summarized below. You can access the PowerPoints submitted by the first three vendors on the Power Users website. The Siemens presentations are posted on the company’s Customer Extranet Portal (CEP). For help in locating them, contact your plant’s service representative.

NEC

Corona in HV stator coils: Theory, causes, repair, and laboratory prognosis

W Howard Moudy, director of operations for National Electric Coil, is a frequent presenter at user group meetings. His goal here was to help plant personnel better understand what corona is, the damage it does, how to identify its presence, and the need to repair the damage it causes early, to avoid the possible need for a coil replacement.

But first, a brief backgrounder on the terminology, extracted from an article written for CCJ by Donald Selkirk, PE, of SaskPower several years ago. The terms corona and partial discharge (PD) are commonly used interchangeably in the electric power industry, he wrote. However, this is not correct.

Both corona and PD are electrical discharges that occur in high-voltage rotating machines (HVRM) when the strength of the applied electric field is great enough to cause ionization.

The IEEE Standards define corona as a “luminous discharge due to ionization of the air surrounding a conductor caused by a voltage gradient exceeding a certain critical value.”

PD is not necessarily luminous or visible. Further, PD may not occur adjacent to an energized conductor and need not occur in air or gas. Additionally, a corona discharge may bridge the entire gap between energized conductors.

Moudy continued: Corona is a surface phenomenon (a “glow”), he said, that leaves its distinguishable mark on winding surfaces, both in the cell and end-winding portions of HVRMs. It is often caused by insufficient clearances between surfaces having different electrical potentials.

Corona also may be attributed to inadequate functioning of the gradient portion of the outer corona protection (OCP) wrap encapsulating stator bars (Fig 1). Erosion of the OCP, he continued, can develop from inside the coil when the ground-wall insulation near the OCP is of poor quality. Mica is a material that may best inhibit corona attack on ground-wall insulation.

The insulation binder is attacked first, followed by the conductive fillers of the OCP material, which may be destroyed. This allows the corona to reach the OCP surface, where it is easy to recognize (Fig 2). Moudy noted that OCP repairs made prior to deterioration deep into the ground-wall insulation are most successful. Given reasonable access, he said, future deterioration in the area of repair can be prevented. Then the prognosis for long-term reliability is very good.

HV generator stator ground insulation repairs

This presentation by National Electric Coil’s W Howard Moudy, director operations, and Gary Slovisky, director of field service, discusses in meaningful detail two case histories—one concerned with the repair of ground-wall insulation, the other with OCP (outer corona protection) repair. The second is a sequel to Moudy’s first presentation (immediately above).

Before committing to a repair, the speakers began, there are things you should know—including the following:

• • • Understand the cause of the failure.
    • Determine the full extent of damage and the suitability of undamaged components.
    • Identify the repair options.
    • Consider the practicality of the repair options identified and their associated risks, and the prognosis in terms of reliability.

Participation and guidance by experienced personnel with a bit of wisdom is essential for a successful outcome, Moudy and Slovisky said. Keep in mind that access and space often are the greatest obstacles in making repairs in the field. Creativity, patience, and experience are essential to overcoming these obstacles.

The root cause of damage to ground-wall insulation in the first case history was a rotor fan blade failure attributed to high-cycle fatigue. One of the 16 fan blades failed near its base but not at the weld joint. The liberated fan tip ring damaged the stator. The generator was relatively small, rated just under 22 MVA. The incident was characterized as a generic OEM design problem. The replacement 13-blade fan was machined from one block of ASTM 4340 alloy steel (no fan tip ring).

Interestingly, some owner/operators of air-cooled condensers equipped with fans having an even number of blades, identified with this problem. Their issues were resolved as well by transitioning to a fan with an odd number of blades.

These three repair options were considered:

• • • Repair failed and damaged coils in-situ and onsite. This offered the shortest delivery time and lowest price, but the lowest evaluated reliability.
    • Factory repair and reinsulate the two failed coils; repair other damaged coils onsite. This option was penalized by a slightly longer delivery time than repair and about twice the cost of repair. However, it was less than one-third the rewind cost and offered a good evaluated reliability.
    • A complete stator rewind was characterized by longest delivery time, highest price, and highest evaluated reliability.

The first option was selected. Presentation illustrates how the ground-wall insulation was repaired successfully, step by step.

The second case history reviews OCP field repairs to a pumped-storage hydro generator and to a generator for an F-class gas turbine. Critical to achieving a successful OCP repair are the following:

• • • Expertise and skilled labor.
    • Use of materials of the highest quality and of proven engineered processes.
    • Assuring an appropriate interface with the ground plane (core).
    • Thorough mixing of semi-conductive and gradient coating treatment materials.
    • Application of the gradient coating in a manner to assure proper/adequate overlap with semi-con coating.
    • Ensure adequate cure time.

MD&A

Preparing generator rotors for cyclic duty

James Joyce, MD&A manager of generator repair operations, set the scene with his opening statement: “The impact of cyclic operation on ac generators is significant. As more units transition from baseload to cyclic duty, the thinking behind generator maintenance and repairs needs to adapt.”

Keep in mind that baseload units operate under creep conditions—that is, constant stress—while cyclic units are challenged by the fluctuating stresses consistent with fatigue conditions.

Major contributors to increased wear on generator rotors in cyclic service, he continued, are these:

• • • Different rates of expansion and contraction of rotor components in close proximity to each other that are made from different materials. Examples include copper coils, insulation, and steel forging (Figs 3 and 4).
    • Expansion and contraction of the retaining rings caused by centrifugal force.

Recall that heating of the rotor winding generally is the limiting feature in generator design. Thus, generators cooled by hydrogen, a more-efficient coolant than air, and are smaller than air-cooled units of similar output and issues related to the expansion and contraction of components are not as severe.

Important to note, Joyce said, that generator manufacturers are under increasing pressure to reduce costs and necessary decisions on conductor cross section, insulation thickness, the amount of steel in the core, etc, negatively impact a generator’s ability to absorb the increased mechanical and electrical stresses of cyclic duty. This puts pressure on O&M personnel to operate their machines within limits suggested by designers to assure the desired levels of reliability and availability can be achieved.

Joyce’s presentation focused on case histories illustrating the effects of cycle duty on the generator fields for a 450-MW hydrogen-cooled unit, a 200-MW hydrogen-cooled unit, a 300-MW air-cooled unit, and an 80-MW air cooled unit. You can learn a great deal from the PowerPoint because the speaker shares the planned work scope for the shop visit, what was found during the inspection, emergent work required, solutions selected, etc.

Highlights include the following:

• • • Cracked blocking required replacement with an enhanced design to prevent crushing and delamination.
    • Replacement of original Nomex turn insulation with a glass laminate material which does not absorb moisture to the extent that Nomex does.
    • Modification of a full-length slot amortisseur, springs, and creepage blocks to accept a top hat pin which locks all three components together, thereby preventing the springs under the slot amortisseurs from migrating during operation.
    • Replacement of original slot armor with Teflon-coated Nomex. Note that the relative movement of the copper coils during load cycling can lead to abrasive damage to the slot armor. The Teflon coating provides a slip plane to mitigate this.
    • The original pole-to-pole connectors were susceptible to low-cycle fatigue as the copper coils expanded and contracted during cyclic operation. They were replaced with an improved design more capable of handling the additional stresses associated with cyclic duty.
    • Copper deformation from cycling dictated coil replacement. Plus, there were broken main leads not normally seen in situations involving high cycles; they too were replaced. New retaining-ring insulation also was installed, the replacement with a Teflon slip plane to prevent the end-winding top turns from deforming while expanding and contracting axially as they heat up and cool down.

Generator findings and case studies

James Joyce’s second presentation provides valuable information for all involved in generator-maintenance decision-making. You’ve heard at user-group meetings about outage extensions to deal with emergent work and may be wondering just how often such incidents occur. Joyce shares MD&A’s experience from the 18 months preceding his presentation Dec 3, 2020 to illustrate how important it is for you to operate your generator within the bounds of the OEM’s recommended practice, review operating data with a keen eye, conduct appropriate tests periodically, and to continually plan for the outage ahead—all this to minimize the number of surprises at the next overhaul.

The manager of generator repair operations said that most fields that come into MD&A’s shop fall into two categories: rewind, or rings-off inspection with a defined scope that does not involve a rewind—such as blocking mods, deep cleaning, or simple testing and inspection.

In the last 18 months, he said, 70% of the units that came into the shop were scheduled ahead of time. The rest were the result of a forced outage. In that latter group, about 12% participated in a field-swap program to minimize outage time. Roughly one-quarter of the scheduled visits required an outage extension because of emergent work (expansion of the original work scope).

Analyzing the root causes of field rewinds, Joyce said 46% of the generators had to be rewound because of a shorted turn or ground fault, 38% because of a main lead or mechanical failure (the latter category includes failed blocking, copper mushrooming, etc), and 12% because of contamination (such as foreign material blocking ventilation passages) that could not be removed. The remaining rewinds were age related.

Joyce concluded his presentation with two case studies, one involving a collector ring failure, the other damage attributed to loss of lube oil. The first resulted from a flashover between the brush rigging and field forging. Once the flashover occurred the collector-ring insulation was compromised, allowing a ground fault from the collector rings to the field forging. Many photos illustrate the extensive damage incurred.

The field winding was unaffected by the flashover. This allowed removal of the damaged section at the end of the rotor and installation of a stub-shaft in its place. Machining of the turbine end journal and oil-deflector surfaces was part of the rework required. Successful inspections and high-speed balance confirmed the field was fit for duty.

The lube-oil failure experienced during startup after a maintenance outage did considerable damage. It occurred because the lube oil to the generator bearings was inadvertently shut off. The incident screamed for attention to detail on the deckplates, the use of checklists, and multiple pairs of eyes on everything.

The resulting bearing failure caused the shaft to drop about 200 mils. The rotating blower hub blades then contacted the stationary housing and shredded, sending the stainless-steel blades throughout the unit. Repair work included machining and weld restoration of journal and seal areas to eliminate hardness, installation of a stub shaft, a rewind and replacement of all insulation components, and blower replacement.

AGT Services

GE GT- and ST-driven generator field repair needs

AGT Services’ Jamie Clark has been on a mission for the last couple of years, presenting at the annual meetings of all major users groups to alert owner/operators about the significant increase in generator failures his company and other service firms are seeing. Clark’s thoughts echoed those of MD&A’s James Joyce in his presentation a week earlier, “Generator findings and case studies” (see above).

Both Clark and Joyce agree that the failures experienced today are related in large part to unit cycling (in particular, those machines designed for baseload service) and age, with lapses in attention to detail during inspection and maintenance contributing.

Clark began his presentation with a chart illustrating the dramatic increase in the number of starts experienced by a combined-cycle plant in Maine over the last decade compared to the start stats for the facility’s first eight years of service. An informal survey that he conducted at a recent meeting revealed a 50% increase in the number of starts among 7F owner/operators. That was an interesting factoid because about half of “planned” projects that AGT Services has been involved in lately have resulted in emergent work. Another 14% of the surveyed users said their starts had doubled, or more than doubled, over the last 10 years.

AGT is seeing more unplanned stator rewinds than ever, Clark continued. Same is true for field rewinds, he added, with some of those resulting from stator failures.

Clark then highlighted the primary areas of the stator affected by cycling, including the following:

• • • End-winding vibration/loosening, noting the higher risk for strand-to-strand series connections.
    • Core-looseness impacts—such as keybar rattle/belly bands and loss of core compression.
    • Slot support system—including wedge system and side packing/ripple springs.

The speaker stressed that all stator parts are designed to work together as a system. Example: Bellybands restrain keybars and when loose allow keybars to “rattle” producing particles of iron oxide. Add in some oil and you have greasing that lubricates the connections, further compromising tightness.

Clark’s extensively illustrated presentation next walks you through the primary field components affected by age/cycling (or design), focusing primarily on GE 7FH2 and 324 generators. Below is a list of the topics he covers. One or more has to be of sufficient interest to warrant a review of slides on the Power Users website. [link]

Slot component migration

• • • Creepage block
    • Amortisseur springs
    • Slot armor deterioration (birth or prior repair defects)
    • Turn insulation

Distance blocking movement

• • • Axial blocks
    • Radial blocks

End turn insulation migration

Copper distortion

• • • Main leads, crossovers
    • Elongation/foreshortening
    • Fat copper in the slots

Braze design/failure

Cracked brazes at corners and on crossovers and leads

Collector systems

• • • Leaking stud seals
    • Collector ring/brush life
    • Field ground detector inoperative

Siemens Energy

Case studies: Cycling impacts on generators

Alejandro Felix, Siemens Energy’s manager of generator service engineering, focuses in large part on the effects of flexible operation on stator end windings and then discusses the OEM’s innovative technologies to address flex-operation needs. For rotor components these include new designs for J straps and pole crossovers. For the stator, new rewind designs are more capable of absorbing axial thermal expansion of the coils. The presentation concludes with a look at Siemens’ service solutions to support flex operation.

Third-harmonic stator ground protection

Tony Cararano, solutions engineer, digs into the details of third-harmonic voltage and how to assure 100% stator ground protection not assured by adhering to IEEE standards. Topics discussed include 59N neutral over-voltage, 27TN third-harmonic neutral under-voltage, and 64S subharmonic voltage injection. Three case studies provide valuable insights.

Hi-pot testing approach

Jim Lau, expert engineer and one of the industry’s most highly regarded electrical engineers, provided a backgrounder on high-voltage testing, covering both ac and dc tests. Dielectric evaluation, preparations to test insulation resistance and polarization index, the value of pass/fail dc hi-pot testing, partial-discharge testing, low-voltage power-factor testing are among the topics reviewed.

In anticipation of the upcoming 2021 virtual conference of the Generator Users Group, readers may find recaps of user and consultant presentations from the 2020 virtual meeting quite helpful. If you are a registered “Power User,” on-demand recordings and slide decks are available here. GUG2021, which operates under the Power Users umbrella, will be conducted on consecutive Thursdays from July 15 through August 5, plus Wednesday, July 21. Registration for the meeting is limited generator owner/operators (users) and comes with no cost.

User and consultant presentations

Use of an ultrasonic device for locating hydrogen leaks

A handheld ultrasonic instrument commonly used for detecting vacuum and air leaks in powerplants is good for locating leaks of hydrogen and other gases as well, reported an experienced user.

The Fluke ii900 Industrial Acoustic Imager (a/k/a sonic leak detector) relies on an array of 64 digital mics to locate the source of the sound (Fig 1) within a frequency band of 2 to 52 kHz. Practically speaking, the instrument can detect a 0.005-cfm leak at 100 psig from up to 33 ft away. It doubles as a camera capable of capturing stills and video, and has a USB-C port for data transfer.

Fig 2 shows the acoustic signal developed by a small hydrogen leak from a generator bushing.

The speaker cautioned that sound can reflect off surrounding surfaces and could be misinterpreted as a leak in the wrong location. If a potential leak source is identified, he recommended viewing the same location from a different angle or distance to verify that the leak source is “true” and not a reflection. Fig 3 is an example of a leak indication caused by sound reflection. The false indication was verified with helium testing.

A thorough understanding of a turbine’s gas piping system and design benefits accuracy. The user added that the default frequency range and filter settings are acceptable for most compressed gas leaks but may require adjustment in noisy environments. This is a trial-and-error “tuning” process.

User safety—avoidance of slips, trips, and falls—was stressed. Surveyors should remain aware of their surroundings while walking and viewing the screen during scans, attendees were told.

Origins of EMI: History and New Research Users Group

The value of radio frequency (RF) for sensing incipient arcing faults in large generators is well known to electrical engineers serving in powerplants. However, questions remain on how to interpret the RF spectrum signature created by the high-frequency currents flowing in the neutral connection. The speaker said there are many possible sources of RF signals—some are within the generator, some external to the generator.

In the first group are the following:

• • • Partial discharges (corona) within the stator-winding insulation.
    • Slot discharges between coil surfaces and the stator iron.
    • Sparking from exciters with brushes.
    • Arcing between adjacent ends of a broken coil strand.

The second group includes the following sources:

• • • Corona and partial discharges in the associated high-voltage power system.
    • Lightning and switching surges.
    • Motors, switches, and other sources in the power station.

Application of the RF arc-sensing technique is straightforward. To measure the complete spectrum, simply clamp an RF current transformer around a generator neutral. Radio noise meters covering the frequency range of interest—about 10 kHz to about 32 MHz—provide the measurements in microvolts quasipeak (uVQP). This is explained as sort of a weighted average approaching the true peak value of the frequency component being measured.

Numerous RF measurements from operating machines have resulted in recognizable RF spectrum signatures, generally repeatable and believed to represent the background levels of normally operating machines free of any arcing condition. Note that preliminary measurements suggest that RF noise external to the generator (refer to short list above) is insignificant.

The speaker called for the formation of a new industry group—perhaps something like one of the users groups covered regularly by CCJ ONsite—to research EMI (electromagnetic interference) signature correlation to a given fault condition, develop tools to interpret the signature to fault conditions, and document and communicate the knowledge worldwide.

One of the tools at the disposal of the proposed research group is an electromagnetic transients analysis program, called ATP, developed by Bonneville Power Authority under a federal grant. The speaker said the program has a significant number of users sharing results globally.

GVPI stator-bar failure root cause, lessons learned

This is a well-illustrated presentation many O&M technicians can learn from. It addresses the failures of two different SGEN6-1000A generators serving gas turbines in a 4 × 1 combined cycle. The four units are characterized by globally vacuum pressure impregnated (GVPI) stator windings (Fig 4).

The first failure was on a 245-MVA, 15-kV machine after nine years of operation. An incorrect cable termination was used during plant construction. The spec called for unshielded cable, but 2/0 shielded cable was used and the shielding was not removed for the approximately 8 in. needed at termination. Because the shielding was not stripped back, it was within strike distance when the fault occurred. The current jumped into the shield rather than travel in the cable conductor, thereby overheating and failing the cable.

The second unit failed a stator-winding hi-pot after 10 years of operation. The test target was 33 kV, 2.2 times rated voltage, as it was for the first unit discussed above. In one phase a bar failed at 30 kV and in another phase a bar failed at 16 kV. Visual inspection showed an “insulation anomaly” on the top surface at core exit on both failed bars. Two other bars that had not failed also displayed the same insulation anomaly.

Several stator bars were extracted for root-cause analysis. A full rewind was performed on this stator. The slides did not comment on the difficulties of removing bars from a GVPI winding.

A CT scan on two failed bars showed signs of what appeared to be insulation cracking internal to the bar at the location of the ridges on both bars that failed the hi-pot, as well as the other two bars with ridges in the insulation. These flaws would be a very serious concern to the fleet of similar units, but the slides did not comment on this issue.

Nonmagnetic retaining ring in-service inspection drivers in 2020 and beyond

In-service inspection of generator retaining rings became an industry standard practice for plant owner/operators in North America and Europe during the mid-1980s, reminded Neil Kilpatrick, principal at GenMet LLC, a respected consultancy on generator metallurgical matters. Many 18Mn5Cr rings, the standard until that time, were found to exhibit significant stress corrosion cracking and most were replaced with 18Mn18Cr rings, which are not susceptible to SCC in water.

But the materials change does not mean you no longer have to perform periodic ring inspections, Kilpatrick said. You never know what might go wrong. He said the following are typical of the failure mechanisms which could cause concerns:

• • • Fatigue fracture (start/stop).
    • Fault-related electrical damage to rings.
    • Fault-related friction damage to rings (rubbing).
    • Subsynchronous oscillation, with torsional fretting and fatigue cracking.

GE 7FH2 extreme vibration during LCI operation

Generator vibration observed during a turbine start using a LCI (load-commutated inverter) was “impossibly” high, the speaker said, showing comprehensive data plots on two slides. Focus of the initial inspection was on bearings (wiped? debris in lube oil?), LCI function, and lift oil (working properly?). LCI was later dropped from the list because it was shared with a sister unit onsite with no issue. GE suggested the cause might be turn shorts, based on its review of the data.

The engineering inspection and evaluation team identified electrolysis at the bearing (Fig 5), significant wear of hydrogen seals, broken shaft grounding brush, and a highly magnetized rotor (upwards of 450 gauss). Bearing repaired, shaft demagnetized to the extent possible with the rotor in place, a restart of the heavily instrumented unit was attempted. Seismic probes revealed an “impossible” 306 ips at 600 rpm.

Flux-probe waveforms showed a coil-to-coil short, with all the turns in one coil and half the turns in another being bypassed. The practical solution given time constraints: a rotor swap. Photos of initial findings when the unit was opened revealed deformed pole-to-pole connectors. Important to this discussion was that the normal pole crossover for a 7FH2 machine was not used when the generator was rewound previously by an alternative OEM and the replacement failed at about 700 starts and 11,500 hours.

Shop work included replacement of the two affected coils using pole-to-pole connectors of the generator OEM’s design. Problem solved.

Conclusions and lessons learned included these:

• • • Things can change completely from one start to another.
    • Confirmed that turn shorts can cause exceptionally high vibration during an LCI start.
    • Reinforced the need to inspect other units in the power generator’s fleet with a similar pole-to-pole connector.
    • If it’s not broken, don’t try to fix it—referring to the change from the OEM’s hairpin pole-to-pole design which has worked well over the years.

SFRA study on generator stator re-wedging

The Sweep Frequency Response Analysis test generally is associated by plant personnel with the physical condition monitoring of transformer windings. It is an efficient way to detect displacement of the transformer core, deformation and displacement of the winding, faulty core grounds, collapse of partial winding, broken or loose clamp connections, shorted circuit turns, open winding conditions, etc.

In this presentation, the speaker presented three case histories and more than 50 slides to show the value of SFRA in determining when stator re-wedging is necessary. There’s still more work to do but the message is clear.

Wonder why wedge tightness is showing up in SFRA data? The speaker explained thusly:

• • • A loose wedging system opens clearances.
    • Clearances permit movement of coils/bars to release installation/migration stresses.
    • Movement opens gaps and contact points affect capacitance and inductive coupling, and resistance to ground.

Things to keep in mind when trying to apply the SFRA data include the following:

• • • Fresh paint affects readings; make sure all paint is cured before gathering data.
    • Meaningful data are limited.
    • The analysis presented is global in nature; local issues may not show.
    • Coil/bar displacement is a dependent variable.
    • The test cases presented are for hydrogen-cooled machines. In-slot partial discharge damage may affect readings for air-cooled generators.
    • The test cases also are for 2-pole machines. It’s unknown at this time if the same patterns apply to 4-pole and hydro units.

AeroPac brushless-exciter flashover

It’s 0800 and the subject unit is synchronized with the grid; power is increased to 110 MW within the hour. Load is raised to 140 MW and the unit trips at 0915 on a loss of generator exciter voltage. The operator’s screen reads, “AVR fault.” No obvious issues are identified and the operators decide to re-energize the unit figuring the trip was “false.” But the unit trips again before it can be synched.

Inspection with assistance from a third-party services provider identified dust on the excitation generator, which was difficult to access. Molten metal was found in the diode wheel; it took three shifts to remove. Decision was made to remove the complete excitation generator housing.

The rotor shaft was damaged during the incident (very deep gouge) so it was pulled and sent to a shop for inspection, analysis, and repair. Electrical test results with the rotor out were satisfactory. Another observation: All six diodes failed but investigators were not able to determine how many failed before the incident. An alarm indicating diode failure was never received. Diodes had never before failed on any of the company’s generators.

Repairs: The portion of the shaft with the deep gouge was removed and a new piece welded it its place. The excitation generator and diode wheel also were scrapped. Shop discovery: A socket head cap screw was found wedged in the diode wheel casing and the connection from the diode wheel to the radial lead was melted in half. Electrical tests received a passing grade; the AVR was eliminated as the root cause.

With the RCA still in progress at the time of the presentation, the plant took the following actions:

• • • Planned to check the tightness of all bolts during every major outage.
    • Purchased a handheld device to monitor diode condition; data would be collected monthly.
    • Initiated work with the OEM on changing the type of filters for the air-cooled excitation generator.
    • Planned to clean the excitation generator every four years and to replace the diodes and their hardware every 10 to 12 years.

7FH2 collector flashover event

This presentation affords the rare opportunity to experience a collector flashover event, which lasted less than an hour, virtually. The generator damaged was a 239-MVA, 18-kV, hydrogen-cooled machine. Data, details in words, graphs of operating data, a dozen photos, etc, are provided.

The outboard collector ring and associated brush-rigging components suffered severe arc damage (Fig 6 left, outboard collector ring is at right in the photo): eight brushes detached completed, seven still attached by their pigtails were free of their holders, nine brushes remained stuck in their holders—attesting to the level of detail provided by the speaker. All 24 brush holders showed arc damage (Fig 6 right).

Repair scope included replacement of the following components:

• • • Entire brush rigging, including new holders and brushes.
    • Both collector rings (the inboard ring could have been reused, however).
    • Outboard collector terminal stud (the existing one had to be drilled out).
    • Seal assemblies on both terminal studs.

Plus, shaft grinding was required to remove harden metal created by arcing.

Identification of the exact root cause of this flashover event was complicated because most of the evidence was vaporized during the incident. Insights gained during the inspection allowed elimination of the following possible causes as unlikely: short brushes, high brush vibration levels, inadequate ventilation, and ambient air contamination.

Among the contributors to this specific collector flashover were believed to be generally low current densities in the brushes, ineffective periodic cleaning of brush holders by contract personnel, poor contact between brush terminals (pigtails) and the outboard collector yoke assembly, and improper orientation of brush holders relative to the collector ring. In brief, the speaker believed the brush holders had outlived their useful lives.

The speaker offered the following characteristics of good collector assembly performance:

• • • Continuous contact of brush to collector ring.
    • Proper brush-to-collector ring contact pressure.
    • Good collector-ring surface film condition.
    • Limited selectivity.

A proven maintenance approach to assure good collector assembly performance focuses on these points:

• • • Routine checks of collector assemblies with rounds on each shift.
    • Weekly checks with the enclosure covers removed—including visual inspection, verification of no abnormal brush vibration, and confirmation of brush freedom of movement within the holders.
    • Monthly, measure brush currents. Compare these to those from the preceding month to identify any obvious current selectivity.
    • Identify and log specific deficiencies, if any, identified with the prescribed maintenance approach.