O&M lessons learned on stator-bar insulation systems help improve generator reliability

Special to CCJ ONsite from Clyde V Maughan, Maughan Generator Consultants

Generator stator-bar insulation systems have experienced a steady evolution since the beginning of central powerplants in the early 1900s. Initially, 1900-1915, natural products were used—for example, shellac, mica flake, paper. These materials apparently worked well on small, low-duty units.

As generators grew in size, problems developed, and during the period 1915-1950 “asphalt-mica” was the common system. But its use was discontinued in the 1950s for generators rated higher than about 25 MW because of materials limitations. Enter thermoset resins, first polyester and then epoxy, for impregnating the mica tapes.

From the 1950s to the present time, thermoset systems have evolved upward in mechanical and electrical capabilities. But during this period, generator ratings also have gone up, with associated mechanical and electrical duties increasing dramatically. Although the still-evolving thermoset insulation systems generally have performed fairly well, problems continue to occur. Some of these problems have been very costly and some have been very persistent. 

From an O&M viewpoint, all thermoset insulation systems used in modern generators are basically the same, although in a few important ways they are different from the soft asphalt systems of yesteryear. Thermoset windings are more vulnerable to cracking of the bar groundwall insulation from sudden short circuits, but more forgiving of over-temperature.  Thermoset bars also are vulnerable to cracking during winding installation.

A few cautions and special considerations relative to the thermoset windings are presented below. But O&M of the soft and hard insulation systems are not greatly different.


Nothing can be seen directly on the winding while the generator is in operation, and monitoring/ instrumentation capability is in general low. However, there are still important operational considerations, including the following:

      • Cleanliness. No insulation system likes contamination. On open, once-through cooling-air systems this can be a huge problem, depending on the local atmosphere. The only solution may be constant physical monitoring of contamination build up.

Even on TEWAC generators, contamination will often slowly build up, depending on tightness of the ventilation system, quality of filtering of the inevitable air ingress, and local atmospheric conditions.

On hydrogen-cooled designs, contamination can still occur, particularly over long periods of time, and primarily during outages. Depending on the nature of the material—for example, coal dust or brush-wear carbon—contamination can be an issue.

Cleanliness issues are greater on the rotor windings, but can be a problem with stator windings as well. It is something that efforts should be expended to minimize, but is likely to continue even under the best of efforts.

      • Moisture is a sub-set of cleanliness. While dirt contamination, and correction, can be well understood, moisture contamination, and its correction, may not be. On open and TEWAC ventilation systems, no prevention really is possible beyond avoiding leaking coolers. On hydrogen-cooled units, humidity can still be an issue with cooler leaks or non-functioning dehumidifier equipment.

      • Direct water-cooled stator winding. These systems require constant monitoring during operation to maintain water purity, correct water oxygen content, water pressure, performance of coolers, and detect water leaks.

      • Overload. System demands may call for overload of a generator. These situations should be limited because excessive load increases the electromagnetic forces on the stator winding as a squared function of current (load).

      • Asynchronous operation. Plant personnel may have little or no control in these situations, except during synchronizing. If the condition occurs at low power, no generator damage may occur. But if at high power, and persisting, total destruction of the generator may result.

      • Sudden short circuit. Except during synchronizing, the operator will have no control relative to short circuits. If synchronizing is off only a few degrees, no damage is likely. If a short occurs at 120 deg, maximum torque occurs and couplings may slip and other damage may occur. If the angle of synchronizing occurs at 120 deg, maximum forces on the stator winding will occur and the stator winding almost certainly will be destroyed.

Winding temperature instrumentation. The recommendation of the OEM should be followed carefully relative to monitoring and responding to winding temperature readings. This may prevent a minor problem from turning into a major, costly, and long forced outage. 

There is an inclination to want to load the unit based on slot RTD/thermocouple (TC) readings. On indirectly cooled windings these readings are very indirect and inaccurate. Read an amalgamation of the temperatures of the cooling gas, the core, and the bar copper through a thick thermal blanket (the ground insulation). 

On direct gas-cooled windings, cooling gas often is measured as discharged from individual bars. These values are an important and reliable indication of winding performance.

On direct water-cooled windings, many designs measure the temperature of the cooling water from each individual bar, and these readings give a valuable monitoring of winding condition.

But on a large number of water-cooled windings, cooling-water temperature is measured as it is discharged from pairs of bars. Using the discharge water temperature and the corresponding slot RTD/TC temperatures, some limited intelligence can be derived as to winding condition. But interpretation of winding condition based on these temperature readings is complex and inaccurate.


A well-designed, properly manufactured, and properly operated generator is unlikely to require rewind in 30 years of operation, and maybe never. But it will require regular maintenance. The frequency of OEM-recommended maintenance has evolved in the last 25 years.

But regardless of maintenance frequency, and considering the discussion in the root-cause section above, some basic principles apply. They are:

1. No work should be attempted without a qualified crew and supervision onsite.

2.The quality of the work and the rate of progress will be expedited if all necessary tools and equipment are on hand.

3. Item 2 also applies to all needed materials and parts.

These three items can be a huge challenge. In particular, skilled workmen are limited in supply and availability. The availability of capable supervision is limited as well. Also, often those sent onsite, even by the best of OEMs and vendors, are not well qualified. The result can be costly in dollars and calendar time, and in quality of work.

In the case of failure root cause determination, if the lead investigative engineer is not highly skilled, incorrect determinations often have been reached and the result have been hugely negative in quality and cost. 

Actual routine maintenance work constitutes a broad spectrum of often highly skilled effort. The work procedures have been documented by the OEMs and by others, and this documentation is broad and voluminous. No attempt has been made here to further document these procedures; but some special considerations are offered below on a few specific topics.

A good inspection by a qualified individual generally is the best assessment tool of a stator winding. It can reveal important information relative to deterioration associated with most of the conditions experienced.

Thus the importance of a skilled inspection cannot be over emphasized. This inspection will consume time, maybe a full shift or more, but it must be done and done by a qualified individual. Many technical papers have been written on this important topic as well as two books with chapters on maintenance, one by this author and the other by Greg Stone.

Test. Some stator-bar deterioration mechanisms cannot be detected by inspection—for example, general deterioration of groundwall, internal PD, strand or bare-bar vibration or displacement, and strand and group shorts. (Turn shorts on a multi-turn coil will normally be detected by winding failure.) Detection of some of these conditions may be possible by available tests, the most important of the tests being over-voltage test (hipot). Many papers have been written on the subject.

      • Hipot is a powerful test, but is controversial because of possible winding failure, which would likely force a long outage for bar or winding replacement.

      • Power-factor test can be performed, although this test has limited usefulness in determining winding condition, in the opinion of some experienced individuals.

      • Partial discharge and electromagnetic interference (EMI) may also provide some intelligence on winding condition. Expert assistance may be needed to interpret test results.

      • Finally there is the low-voltage insulation resistance test—the “megger” test. Both resistance and polarization index readings should be taken at every convenient opportunity.

Other maintenance considerations:

      • Robotic inspection without rotor removal is widely recommended by OEMs, and these devices can perform rather well in an inspection of stator core areas. They can also do wedge tightness test and ElCid core insulation integrity test. However, a robotic inspection is expensive and may indicate the need to remove the rotor anyway.

      • Re-wedging of the stator winding is recommended where it is not needed. Judgement of wedge tightness can be very subjective—for example, manual test by inexpert individual, improper use of tightness test device, misunderstanding results of tightness device, etc. Re-wedging is expensive and time consuming, and can result in core and/or winding damage. Also, if only the end wedges are loose, only the end wedges should be replaced, not all wedges.

It is essential that a qualified expert is involved in any wedge tightness assessment decision.

Posted in 501 F&G Users Group |

Comments are closed.