This is the second and final group of presentations by owner/operators at the second annual meeting of the Generator Users Group (GUG), today organized under the Power Users Group umbrella. The GUG meeting was conducted in parallel with the annual conferences of the Combined Cycle Users Group (CCUG) and Steam Turbine Users Group (STUG) in San Antonio, Aug 22-25, 2016.
The summaries below were compiled by Clyde V Maughan, president, Maughan Generator Consultants, the muscle behind the launch of the GUG in 2015. Presentations by consultants and suppliers, also profiled by Maughan, are aggregated separately. If you missed the first part of CCJ ONsite’s coverage of the San Antonio meeting, it’s only one click away.
User presentations made at the GUG’s second annual meeting and summarized in this issue are listed below; links provided enable quick access to the topics of greatest interest to you.
Interpretation and correlation of EL CID results to rated flux testing, Ryan Harrison, P Eng, ATCO Power.
Continuous EMI monitoring, Kent Smith, Duke Energy.
Generator overheating phase-to-phase failure, Craig Spencer, Calpine Corp.
Handley Generating Station Unit 4 excitation failure, Joe Riebau, Exelon Power.
Experience with Alstom air-cooled generators—Part II (stator phase connection conductor fatigue), Jeff Phelps, Southern Company.
Emergency field rewind, Fred King, Duke Energy.
SFC flashover at Mystic Generating Station, Kapil Inamdar and Joe Ribeau, Exelon Power.
Users wanting to dig deeper into one or more of these areas can access the presentations of interest on the Power Users website. You must be registered to participate in the forum, a relatively simple process if you’re not already signed up.
Interpretation and correlation of EL CID results to rated flux testing
The generator discussed in this presentation is a nominal 400-MVA, hydrogen direct-water-cooled unit which was installed in a coal-fired plant (COD 1980). The stator was fully rewound in 2004, during which time a significant core fault (287-mA peak value) was discovered by EL CID test after the new winding was installed. Subsequent rated-flux testing and progressive EL CID flux testing confirmed the concern. Part of the winding was removed and a slot-bottom section repaired (with no root cause determined). After repairs, the core was left with EL CID 65-mA peak.
Top-tooth burning and progressive thermal artifacts have been monitored during subsequent inspections with hot spots greater than 10 deg C noted; however, no EL CID readings exceeded 100 mA. The conditions on this core prompted the following questions:
Does trending the EL CID results offer meaningful condition trending information?
Do the EL CID results correlate to the rated-flux test results?
Investigations were conducted to answer these questions, with results discussed by the presenter.
After digitizing select EL CID readings, early minor EL CID indications were corrected to rated-flux hot-spot data with the following mixed results:
In most cases, EL CID signature correlates to areas where a hot spot exists, but not always.
Both tests have inconsistencies in results across time periods.
EL CID was judged “more repeatable,” provided the same excitation is used.
More-frequent EL CID testing is preferred over less-frequent rated-flux testing.
Comparisons were made of various data: pre-rewind, post winding removal, post high-flux testing, and post repair and high-flux test. With adjustments, EL CID historical data were fairly consistent for trending progressive degradation of the core. Attempts were made to correlate high-flux and EL CID data, with mixed results.
At this point, some fundamental questions remain—for instance:
At what EL CID value should additional steps be taken—is a value below 100 mA appropriate in some cases?
Is it possible that rated-flux testing can initiate or advance existing damage deep within the core?
Understanding the detail of core-flux testing remains something of a mystery because results cannot always be taken at face value. It appears there is great opportunity for benefit related to efforts such as discussed by the presenter. For example, he is developing digital tools to facilitate fast and easy trending of EL CID results. These will be shared at a later date and users should find them extremely valuable. An update on progress is likely at the 2017 GUG meeting in Chandler (Phoenix), Ariz, Aug 28-31.
Continuous EMI monitoring
Motors, generators, transformers, and switchgears typically are monitored with hand-held instrumentation. One informal survey found about half of the nation’s powerplants use portable EMI (electromagnetic interference) monitors. EMI also can be tracked with a radio-frequency current transformer (RFCT) placed, for example, on the generator grounding cable (Fig 1).
The speaker told generator users that a new input card had been developed by National Instruments to monitor continuously the output from the RFCT. Duke Energy currently has this instrumentation on 72 transformers and 53 generators at 15 sites. Capabilities of the diagnostic equipment were said to be considerable: full-spectrum scan, live-frequency visual, live-frequency audio, historical-spectrum viewing, power-spectrum trending—five bands with remote access from anywhere on the Duke Intranet.
Fig 2 compares two full-spectrum plots of sister units. The red trace is for a unit that had significant vibration problems, with multiple plant trips from secondary CT wiring being cut. The CT wiring has been stabilized but this trace still shows significant electrical activity. The blue trace is the sister unit at the same location and reveals no signs of any major issues.
The full-spectrum plot in Fig 3 compares scans of the same unit taken approximately two months apart. It shows electrical activity increased slightly over time. The fact that there is significant activity in the high-frequency areas leads engineers to believe there’s also significant electrical activity near the isolated phase bus (IPB).
In addition to the full-spectrum scan, plant personnel took local measurements with an EMI “sniffer.” It also indicated significant electrical activity in the IPB area. Local measurements show high EMI levels in the bushing-to-bus transition area, as well as in the potential-transformer area. Plans are in place to inspect these areas during the spring 2017 outage.
The diagnostic systems discussed were said to provide plant personnel valuable equipment condition information; interpretation of this information will become better as more experience is gained.
Generator overheating phase-to-phase failure
The subject 300-MVA, 18-kV generator was manufactured in 2000 and installed in a combined-cycle plant. On Feb 27, 2015, the plant was removed from service to upgrade the steam turbine/generator’s DCS. The change involved converting from the steamer OEM’s DCS to one installed by a different OEM on the gas turbines.
The plant returned to service Mar 16 at 10:10 a.m. Six hours later, the steam turbine tripped on relays 27, 86-1, and 87G. Both gas turbines tripped as well. The initial walk-down of the steamer revealed smoke coming out of the exciter-end generator bearing cavity; the generator frame was too hot to touch anywhere. Data analysis revealed a peak instantaneous fault current of 74 kA.
Generator cooling-water flow was controlled with a throttle valve; the throttle-valve’s position was controlled by logic, with inputs from stator hot-gas RTDs. The hot-gas RTDs were accidentally configured as “J”-type thermocouples at the ADC signal processing card during the DCS upgrade, so the “measured” temperature never exceeded 27C, thus leaving the valve throttled “off.” Back calculated, the generator gas temperature actually had reached an estimated 213C.
The DCS malfunction caused gross overheating of the generator. Damage was severe during the four-hour operating period (photos left and center), which led to the connection-ring failure shown at the right in the photo array.
Copper splices were applied to the connection ring and phase dropper, and were reinsulated locally. The stator and coolers were thoroughly cleaned and stator endwindings were treated with wicking resin. The core was requalified by both EL CID and loop/ring testing. The rotor exterior was cleaned.
The unit returned to service May 2. A stator rewind kit, purchased as a contingency, will be installed on a planned basis at the next turbine major outage—or sooner.
Handley Generating Station Unit 4 excitation failure
Handley 4 is an Allis Chalmers hydrogen- and water-cooled generator with brushless excitation. It went into service in 1976 and is used today primarily in peaking service. The main exciter is rated at 600 V/4500 amps. Conversion to DC is accomplished through a rotating 3-phase rectifier comprised of inboard and outboard diode wheels, each with eight fuses, eight heat sinks, and 16 diodes per phase.
Handley 4 was called upon to perform a required reactive capability test which dictates unit operation at maximum megawatt output and at maximum lagging MVARs for 15 minutes. The test was to be completed under a recently developed procedure for reactive capability testing. During ramp-up from no-load, the generator briefly exceeded the published maximum excitation limit. It was quickly brought back within the machine capability curve.
The MVAR output was above historical levels, but all generator parameters were acceptable and within manufacturer limits. Approximately 12 minutes into the test, the over-excitation limiter and instantaneous limiter alarms were received and the unit tripped from service.
Subsequent investigations identified significant damage to the diode wheels. All fuses on the inboard wheel were found open. Severe damage was noted on two of the diodes. Both of these fuses were associated with the same phase and had completely blown apart (Fig 1).
Parts from the failed fuses were ejected, damaging remaining components in the wheel as well as the wheel itself (Fig 2). Severe heating and arcing damage was noted on two heat sinks and their associated insulation.
Repairs required complete disassembly of both the inboard and outboard diode wheels. All fuses and diodes were replaced, damage to the inboard wheel was repaired and NDE tested, and heat-sink insulation and the two damaged heat sinks were replaced.
During the repair process, components of the undamaged outboard diode wheel were electrically tested. A large number of fuses were found open-circuited—including five out of eight fuses on one phase. Although each of these fuses is equipped with a fuse-failure pop-up indicator, none of the indicators activated. (Pop-up indicators only activate if the fuse fails electrically.) The owner’s engineers concluded that fatigue failure of the fuse elements caused the open circuits. Periodic testing of fuses is necessary to ensure their integrity.
Further investigations concluded that only two of eight fuses were in service on one phase at the time of the failure. This caused overloading of the two circuits, and subsequent overheating of two heat sinks. The insulation under the heat sinks burned and allowed electrical tracking and a phase-to-phase fault in the inboard diode wheel. The surge in fault current caused the two remaining fuses to blow apart, and the unit tripped from service.
Actions to prevent future incidents include daily inspections of the diode wheel, increased testing of diode-wheel components, revision to fleet reactive capability test procedures, and improved operator training.
Experience with Alstom air-cooled generators—Part II
Initial awareness of the phase-connection issue discussed by the speaker came in August 2010 following a stator-winding in-service failure. Root cause: phase-bar resonance. The failed unit was repaired and returned to service.
An aggressive plan implemented for the company’s seven air-cooled generators of the type described focused on inspection and repair of “at risk” units. It called for repairing damaged strands (where necessary) and improving the support blocking scheme at phase bars. The blocking-scheme mod has evolved and periodic maintenance is anticipated, including periodic natural-frequency testing. A monitoring program is in place to avoid a repeat failure event.
A winding in satisfactory condition is shown in Fig 1 (left). A close-up of the area is alongside. The failed joint is shown in Fig 2. The extensive burn damage resulted from the arc which continued to carry current for several seconds after ground relay trip as the field current decayed.
Fig 3 (left) revealed cracked strands (at the tip of the pen), which are shown close-up at right. Fig 4 shows the repaired connection with additional blocking and tying.
The Alstom design places a resistance temperature detector (RTD) in each of the 12 phase-lead slots. Operational data show that the RTDs in slots with many cracked strands clearly exhibited a higher temperature rise than all remaining RTDs prior to cracked-strand repair. Data mined following the repairs show the temperature rise of each RTD returned to values consistent with the overall average of all 12 stator RTDs. Thus careful attention to RTD readings offers an opportunity to remove a unit from service before winding failure.
Emergency field rewind
Numerous maintenance problems with generators in the Duke system were described with slides narrated by the utility’s Fred King and AGTServices Inc’s Jamie Clark (access the presentation for more excellent photography). Issues included broken J-straps (Fig 1). On another unit, a flux-probe test revealed shorted turns in a large coil. Inspection revealed the root cause as movement of turn insulation (Fig 2).
Failure of an exciter lead is shown in Fig 3 (left) with the upgraded connector to its right. Several cases of endwinding and connection-ring vibration have been experienced by Duke generators with indications as seen in Fig 4. Each of these was corrected by tie replacement and/or application of bonding resin (Fig 5).
Answers to several informal industry survey questions were provided by the presenters for everyone’s benefit. The percentages of “yes” responses follow the questions below:
Have you experienced J-strap failures? 50%
Do you require a pressure test on bore seals on hydrogen-cooled units? 89%
Have you found field slot-liner problems requiring field rewind? 53%
Have you operated a unit with one field ground? 47%
Do you require new copper for field rewinds? 5%
Do you require a high-speed balance after field rewind? 70%
Do you specify stator wedge materials for rewedge/rewind projects? 53%
SFC flashover at Mystic Generating Station
Mystic station consists of eight generating units, six of which are arranged in two separate 2 × 1 combined-cycle blocks. The gas turbines require use of a static frequency converter (SFC) for startup. A precise start-up procedure is followed, one using multiple buses ranging from the 5-kV SFC output circuit to the 16-kV-rated generator bus.
Startup consists of operating the SFC and excitation concurrently to bring the units up to 2400 rpm. The SFC is then switched off and disconnected, and combustion takes the units up to 3600 rpm. Finally, excitation is reapplied near 3600 rpm as the units become ready for synchronization to the grid.
During a troublesome start of one gas turbine, the SFC circuit failed to disconnect from the generator bus. This led to a direct connection between the SFC and generator after 2400 rpm. Then excitation was reapplied at 3600 rpm and voltage increased to 16 kV. The application of 16 kV on the 5-kV-rated SFC circuit led to failure of the SFC and caused multiple cable failures in trays linking the SFC to the generator bus.
A subsequent investigation determined the cause of the event as failure of the SFC disconnect switch to remain open after 2400 rpm. Insufficient noise filtering in the plant DCS and absence of feedback loop between field breaker and SFC disconnect switch position were deemed contributing factors.
Corrective actions implemented consisted of protection logic modification to add interlocks between SFC disconnect switches and field breaker, as well as between the SFC disconnect switches and generator neutral ground disconnect switch. Dead-band filters also were installed in the plant DCS to improve noise filtering, and field-breaker trip logic was modified to achieve faster tripping.
Temporary conduit running from the length of the generator bus to the second plant SFC was installed to reach plant operational availability within one week, but complete repair to original condition took significantly longer and required OEM involvement.
Kapil Inamdar is an engineer on the central engineering staff responsible for providing technical support to Exelon powerplants; Joe Riebau is senior manager of electrical engineering