Stator winding problems

Mike Bresney, AGTServices

Bresney began his presentation by showing photos of generator failures attributed to stator end-winding vibration, such as that in Fig 1. Next, he discussed criteria to prevent serious end-winding vibration—specifically, avoid the following:

• Resonance in the range of +10% to -5% of driving frequency—or 132 to 114 Hz on a 60-Hz machine.

• Relatively low response amplitude.

• End winding responding as an integral system.

An example of a poorly integrated end-winding support system was presented, along with modifications made to fully correct the problem. Then Bresney discussed options for in-situ repairs of water-cooled stator bars that had suffered cracked strands away from the water-box braze, somewhere within the bar itself. Repair options were explained that would return the existing bar to reliable service versus replacing the bar, as typically recommended. Access presentation.

Partial-discharge problems

Ryan Harrison, P Eng, ATCO Power (Canada) and vice chair, GUG steering committee

ATCO installed six generators (nominal 100 MVA/13.8 kV) at three plants in 2003 and they have shown heavy indications of surface partial-discharge (PD) attack (Fig 2). No failures have occurred in these windings to date, but vendors have offered to perform “repairs”—that is, a temporary fix to reduce PD at sites where the solution is applied in the end windings. Hipot also was recommended but not performed pending purchase and delivery of a spare rewind kit.

Actions taken by the owner included the following: purchase of a spare winding, installation of PD monitoring capability, crawl-through inspections when possible, participation in the Generator Users Group, and third-party consultation.

Readings from the PD instrumentation gave somewhat confusing results. Plant personnel found that “repairs” can make things worse, the huge amount of data collected can be difficult to summarize, specific conditions at different plants may influence PD production, and the predominance of positive data suggested surface PD (as expected).

One unit then was “repaired,” with pre- and post-maintenance DC hipot, without failure. A second unit was rewound. Before removal of the damaged winding, a hipot test at 2E+3 was conducted and did not fail.

The plan going forward is to keep on hand a spare winding, perform diagnostics on bars from the windings removed, attempt to understand the impact of deferring rewinds past the next major inspection, and rewind based on site risk and opportunity. Access presentation.

Electromagnetic interference testing

Jim Timperley, P Eng, Doble Engineering Co

Timperley introduced the group to two methods for detecting EMI:

• Radio-frequency (RF) current transformer reading the frequency spectrum of the ground current for the machine being inspected—usually a motor or generator.

• A small hand-held instrument, the so-called “sniffer,” to measure radiated energy from any powerplant component.

For the former, obtaining the signal is not difficult, but interpretation of the output is complicated and may require considerable technical background and training. By contrast, the “sniffer” test is easy to perform and the output easily interpreted.

Timperley told the group RF energy is conducted and radiated from every PD or arcing event location. The “sniffer” in Fig 3 detects and measures the strength of this energy in operating equipment, providing guidance on where to look for problems. The technique is fast and safe. No doors are opened, no flash hazard, no connections made to energized systems.

The “sniffer” has been used for years to locate, and correct before failure, many potentially serious problems with generators, excitation equipment, motors, transformers, switchgear, bus, bearings, and other equipment. Access presentation.

Partial-discharge testing of stator windings

Greg Stone, Iris Power, a Qualitrol company

Dr Stone explained that PD is a symptom, or cause, of several winding deterioration processes—specifically, bar vibration in the slot (Fig 4), insulation delamination and tracking in contaminants (Fig 5), and ground-wall voids and inadequate spacing between bars in the end windings (Fig 6).

Attempts to use PD for the detection of stator winding problems date back to the early 1950s, but problems of electrical interference resulted in the need for experienced experts to interpret the data. As late as the late 1980s, an EPRI program manager called online PD testing of stators “witchcraft”—and many utilities agreed.

However, much progress was being made in noise separation, sensor reliability, signal interpretation, and relationships between PD signal and winding condition. Today, after six decades of work in technology development, PD monitoring generally is regarded as a standard method for condition assessment of stator windings, with instrumentation installed on over 15,000 generators and motors. Access presentation.

Copper oxide issues in water-cooled stator windings

Matthias Svoboda, SvoBaTech AG (Switzerland)

Water cooling of stators has been used since the 1950s for large generators, introducing several problems in the process—including the build-up of copper oxide on heat-transfer surfaces (Fig 7). Svoboda’s presentation focused on detection of oxide formation and methods for its removal and prevention.

The ability to monitor heat exchangers for oxide formation is limited, he said. No direct method is available. However, indirect monitoring is possible. Example: Migrating oxides caught in filters, if excessive, are an indicator of possible problems. Use of thermocouples to monitor water temperature from individual bars is another indirect indicator. But the reliability of this method is questionable for designs with flows combined from pairs of bars. Use of slot RTDs generally is not recommended.

Copper in water is prone to oxidation but the reaction proceeds slowly at low concentrations of oxygen. Intermediate concentrations of oxygen are problematic, as shown in Fig 8. Note, also, the impact of pH: as it increases, oxide releases decrease except for intermediate concentrations of oxygen. Svoboda’s experience indicates that pH control has both advantages and disadvantages.

Removal of oxide buildup can be challenging. Chemical cleaning is reliable if the accumulation is not severe. But if passages are completely plugged, chemical methods by themselves are ineffective and a combination of mechanical and chemical cleaning is required. This is time-consuming and expensive, and satisfaction is not guaranteed.

Chemical cleaning itself can be relatively simple and may be done online. If cooling passages are not blocked and the process is done correctly, all build-up can be removed from the system. Cleaning with ethylenediaminetetraacetic acid (EDTA) generally is recommended. It removes only copper oxides, not other materials in the system—such as silver brazing. But expert know-how in the use of EDTA is recommended to minimize risks and assure effectiveness. Access presentation.

Stator winding tests

Bert Milano, consultant

The noted consultant began his presentation with a review of general AC and DC test procedures followed by an explanation of engineering principles behind the polarization index (PI) test. With this understanding it becomes clear that considerably more information can be obtained by plotting the data rather than simply taking readings in 1- or 10-min intervals.

High-potential testing was discussed in detail. Two methods are used: DC and power-frequency AC. The latter is a go/no-go test. If the winding fails to hold the voltage, repair will be necessary. As a result, many utilities will not perform this test. But a winding that passes with an appropriate test voltage is likely to be safe for operation another three to five years.

Milano discussed three types of DC tests: fixed-increment, time-graded, and ramp. Each may allow interruption of the test prior to winding failure, and for this reason (and the small, light test set used) many owner/operators prefer the DC test. The speaker focused on the ramp test, discussing its advantages and providing examples of its use. One company has used this test for more than 30 years, he said, and has completed more than 5400 tests. Of those, only four sudden failures occurred—that is, the test was not interrupted before the winding failed. Access presentation.