Busy generator major: Three failures identified, corrected

Generator problems can be challenging to solve given the complexities of equipment design, difficulties in identifying the true root causes of failures, proper evaluation of alternative corrective actions and of alternative contractors, etc.

The following case history, shared by Relu Ilie, Israel Electric Corp’s (IEC) engineer in charge of powerplant electrical equipment, describes three significant issues found and addressed during the major inspection of a 35-year-old, 2-pole, 269-MVA generator. This outage was conducted five years ago and the repairs described here have prevented a recurrence of these failures since that time.

The generator is a hydrogen-cooled unit with a directly cooled stator winding using the water-box design shown in Fig 1. Each water-box segment supplies the nozzles serving several winding bars as well as the series bar-to-bar electrical connectors.

Generator Failure Figs 1

Stator bar failure. For a couple of months before the major inspection, plant personnel recognized hydrogen was leaking into the cooling water at a continuously increasing rate. During the outage, the leak was located inside one of the water boxes at the turbine end.

After dismantling the water box and bar insulation, a broken bottom bar was found. Most of the 18 sub-conductors were found broken; only two or three strands kept the water nozzle in place (Figs 2 and 3). The bar still was able to carry current and the broken strands were still closed enough to allow some cooling-water flow.

Generator Failure Figs 2, 3

It did not appear that there was any significant leakage of water out of the broken strands into the stator-bar insulation, which is why a ground fault did not occur. Note that while hydrogen pressure is higher than water pressure, water still can leak out of the strands by capillary action.

IEC engineers were well aware that replacement of a bottom bar would entail the removal and replacement of approximately 28 top conductor bars—an outage-duration concern. The OEM proposed an interim possible repair: Cut out the damaged section of the conductor bar and braze on its end an extended nozzle—about 3 inches—to revert the conductor bar back to its design length, in-situ. Management accepted this repair plan. After its implementation, the individual bar and the complete stator winding were fully leak-tested (pressure and vacuum) before the machine was returned to service.

Loose rotor-wedge locking screws. The rotor was rewound eight years before this major outage by a highly regarded non-OEM supplier. In the center of the rewound rotor there are 28 1-in.-long screws, each located at the longitudinal midpoint of the tooth adjacent to every active slot.

Inspection of the screws revealed the following: One was found partially unscrewed, another could be unscrewed by hand, and three were completely missing (Fig 4). Fragments of broken screws were found in the blower housing at the turbine end.

Generator Failure Figs 4-6

Utility personnel identified thousands of points of impact damage on the rotor surface along its full length. This was attributed to rotor-wedge locking screws that came loose and rattled around the stator bore (Figs 5 and 6). A positive: The retaining rings were not damaged—the loose screws probably stopped by the gas baffles or held by the magnetic field.

After the fault, it became clear that the rewind contractor probably applied its standard design of full-length wedge locking against axial movement, by use of central screws located at wedge side. This method is intended, in an axial-cooled generator design, to address the potential issue of ventilation-hole blockage, and also to avoid wedge-to-retaining ring contact and associated arcing.

However, the IEC machine was designed originally for rotor radial cooling with wedges allowed to move (slightly) axially, independent of the windings or rotor body. Slot-wedge screws were not used by the OEM because of the risk of fatigue cracks either in the wedge or rotor tooth.

Fortunately, damage to the rotor forging and wedges was minimal and inconsequential. Evidently, the rewind contractor didn’t pay enough attention to securing the screws. As corrective action, mechanical locking by staking was employed, in addition to suitable Loctite bond.

Stator core damage. The stator also was damaged seriously by the screws that liberated from the rotor. Hundreds of points of significant damage were found along the entire length of iron core surface and on wedges. Additionally, tens of core areas were found smooth/polished, meaning a huge number of severely shorted laminations, including excessive heating and core iron melting on the tooth tips in some areas (Figs 7-9).

Generator Failure Figs 7-9

A stator-core replacement was out of the question, because of the time and cost involved. When the bottom-bar damage described earlier was repaired, stator-core damage locations were ground, using a high-speed grinding wheel, and electro-etched. This required several weeks of difficult work done with patience.

The repairs were tested and confirmed by EL-CID tests and high-flux (loop) tests with infrared scan. In some locations, follow-up grinding and electro-etching were necessary and the tests had to be repeated several times.

Failure root cause. Two strong and relatively long short-circuit events occurred three and four years before the inspection. These events were close to the generator: One on the HV bus of the neighboring substation and the second on main transformer HV terminals.

IEC engineers suspected there was significant cumulative effect of the end-winding movement that caused crack initiation in the copper strands. After that, vibration in operation likely fatigued the copper and caused the rest of the damage.

Conclusions. From the failures described above concerns are raised relative to the use of OEM versus non-OEM shops to perform critical refurbishment work—such as rewinds.

The described events illustrate one worthwhile and creative solution performed for the stator-bar failure by the OEM, and, on the other hand, a deficient rotor rewind done by an alternative service provider. Of course, in other cases, the results from using the OEM versus an alternative supplier may be completely different.

A second conclusion: Major refurbishments which are supposed to eliminate existing problems, may also introduce new weaknesses and lead to other types of failures. Further maintenance difficulties may result when main generator components (rotor and stator, for example) are made/rebuilt by different suppliers.

Using supplier qualification process for reverse engineering work may help; however, it cannot solve problems related to a lack of deep understanding of the OEM’s design. Another possible help may come from supervising the entire process by knowledgeable in-house specialists or independent highly-qualified consultants.

Posted in 7F Users Group |

Comments are closed.

Categories