Issue1_06 Example of Premature Stator Winding Failure in Recentrly Manufactured Motors and Generators

7/29/2019 Issue1_06 Example of Premature Stator Winding Failure in Recentrly Manufactured Motors and Generators

http://slidepdf.com/reader/full/issue106-example-of-premature-stator-winding-failure-in-recentrly-manufactured 1/4Electricity Today24

Local manufacturers of motorsand generators must nowcompete with other manufac-

turers from around the world. The resulthas been pressure on the manufacturersto reduce costs through process..Improvements, using new materialsand/or increasing design stresses.

The stator winding of air-cooledmotors and generators has been a partic-ular focus for cost reduction or converse-ly rated power increases in the same sta-tor frame inevitably, the adoption of these new approaches has resulted insome problems and even failures, some-times just a few years after the machine iscommissioned.

This article provides examples of premature stator winding insulationaging problems, including failurescaused by poorly applied PD protection

layers, coils installed too close togetherin the endwinding,coils that abraded dueto inadequate bracingin the slot as well asproblems associatedwith the global VPIprocess. This anecdo-tal information aboutthe perceived increasein the premature fail-ure rate is validatedby a statistical analy-

sis of on-line partialdischarge data thatshows that somebrands have seen alarge increase of PDin machines made inthe past 10 years.Some suggestions aremade that may enablemachine owners toreduce the risk of pre-mature winding fail-ure.

INTRODUCTIONAs part of the analysis of on-line

partial discharge (PD) data performed onthousands of motors and generators, ithas been noted that stators made by somemanufacturers in the past decade havemuch higher PD than stators made by thesame manufacturer more than 10 yearsago [1]. For example, Figure 1 shows thepeak PD activity vs. winding manufac-turing date for 9 of the world’s largestmanufacturers of air-cooled motors andgenerators. This figure shows that for on-line PD readings measured in 2003, fourmanufacturers are exhibiting much high-er PD on recently made stators, than theytypically experienced on their machinesmade before 1995. Since high PD can beoften associated with rapid aging of thestator winding insulation system, thehigh PD in recently manufactured stators

is of concern.

Maughan has recently published acatalog of premature winding problemshe has seen [2]. This paper presents addi-tional examples of advanced insulationaging that this author has recently seen inair-cooled motor and generator statorsMany of these deterioration processesresulted in premature winding failureThis paper also postulates reasons whysome recently made stator winding insu-lation systems have higher PD activityand some steps the user may take toreduce the risk of premature stator winding failure.

ELECTRIC STRESS CONTROL COATING PROBLEMS

Most stator windings rated >6 kVemploy a carbon-loaded paint or tape onthe surface of the coils or bars in the slot[3]. This ‘semicon’ coating prevents PDbetween the surface of the coil and the

stator core in any small air gap tha

GENERATION

EXAMPLES OF PREMATURE STATOR WINDINGFAILURE IN RECENTLY MANUFACUTRED

MOTORS AND GENERATORSBy G.C. Stone, Iris Power Engineering

Figure 1: Peak PD activity versus year of manufacture for 9 major manufacturers


http://slidepdf.com/reader/full/issue106-example-of-premature-stator-winding-failure-in-recentrly-manufactured 2/4

inevitably exits at this interface. In addi-tion, most manufacturers use a siliconcarbide-loaded paint or tape on the coilfor 10 cm or so outside of the slot. Thissilicon carbide coating overlaps thesemicon coating, and reduces the highelectric field that would otherwise existat the end of the semicon coatings.

In the 1970s, there were a number of machines that exhibited very high PD

and high ozone concentrations fromeither or both coatings that were causedby manufacturing problems. The prob-lems seemed to originate from coatingswhere the carbon and/or silicon carbidewas non-uniformly dispersed in the insu-lation matrix or where the applicationmethod resulted in microvoids just underthe coating. In both cases the result wasPD. This high PD created ozone thatchemically attacked the insulation (notto mention heat exchanger metal andrubber components) and properly madeareas of the coatings, resulting in the

spread of the problem. This problemseems to be worse if the winding insula-

tion operates at higher electric stressand/or higher temperature. Perhaps it isfor this reason that there seems to be arecurrence of this problem in the pastfew years.

Figure 2 shows a hydrogeneratorstator where a very noticeable whiteband is visible at the junction of thesemicon coating and the silicon carbidecoating. Figure 3 shows a winding where

the semicon has virtually disappeared inthe slot due to poor application of thesemicon coating.

ENDWINDING PD

Coils operating at high voltage and

placed adjacent to other high voltagecoils in another phase require a mini-mum separation to avoid PD in the airspace between coils. This PD will gradu-ally erode the groundwall insulation andmay lead to phase-to-phase stator failure.The higher the voltage class of themachine and the thinner the groundwallinsulation, the greater must be the spac-ing [3].

Unfortunately, in many motors andgenerators we have noted inadequatespacing, and consequently high PD (and

ozone). Figure 4 shows the white residuecaused by ozone resulting from PDbetween two coils in different phasesthat were installed too close to one-another.

Figure 5 shows a stator where theclose spacing has resulted in highlystressed air adjacent to where two phaseend coils (in different phases) make con-nection to the circuit ring bus. Theresulting partial discharge eventually

January 2006

Continued on Page 26

Figure 2: Semicon and grading coatingoverlap deterioration due to poorly applied or inadequate coatings

Figure 3: Destruction of the coil semiconcoating in the stator slot due to PD andozone

Figure 4: PD occurring between high volt-age coils in two different phases, where thespacing is insufficient to prevent the PD.Note that the spacing between adjacent coils is irregular, as a result of poor manu-facturing processes.


http://slidepdf.com/reader/full/issue106-example-of-premature-stator-winding-failure-in-recentrly-manufactured 3/4Electricity Today26

bores a hole through the insulation, trig-

gering a phase-to-phase or a phase-to-ground fault.

LOOSE COILS IN THE SLOT

Coil vibration in the slot has longbeen a problem in all non-global VPI sta-tors made with thermoset insulation sys-tems such as epoxy mica. The firstinstances were reported over 50 yearsago [3]. The root cause of the problem isthat at full load, the twice power frequen-cy magnetic forces will vibrate the coilsif the coils are not tightly held in the slot.

Consequently, the groundwall insulationrubs against the laminated steel core – avery abrasive surface. First the semicon-ductive layer of the bar or coil is abradedaway, and then the groundwall insula-tion. The mechanism is sometimes

referred to as slot discharge,because once the semiconduc-tive coating is abraded, partialdischarges occur between thecoil surface and the core, fur-ther increasing the rate of dete-rioration.

Figure 6 shows a coil thatwas removed from the slot of a

hydrogenerator where thesemicon and about 30% of thegroundwall thickness has beenabraded away. The manufac-turer had not mechanicallysecured the coils in the slots bymeans such as sidepacking,ripple springs, two partwedges, conformable restraintin the slots, etc. Figure 7 showsa bar in the process of beingremoved from the stator slot that showssignificant abrasion of the insulation inthe slot, for the same reasons.

Normally one would not expectloose coils to be a problem in a globalVPI stator, since the coils are effectivelyglued to the stator core. However, if thecoils are made too small for the slot, andare subjected to load cycling that createsshear stresses between the coils and thestator core, then loose coils and slot dis-charge may occur in some designs (Fig-ure 8).

AVOIDING PREMATURE

STATOR FAILURE

The premature failures describedabove were a consequence of the designand/or manufacture of the stator. Specifi-cally:

• The electric stress control prob-lems may be caused by poorly appliedcoatings. The deterioration process isaccelerated in winding designs that causea class F insulation system to operateabove about 120C and/or with an averagegroundwall electric stress above 3kV/mm.

• The endwinding PD is probably

caused either by: (a) poor dimensionalcontrol of the coil and/or inconsistentalignment of adjacent coils in the slots;(b) too short an endwinding which doesnot allow enough circumference at thecoil ends for sufficient air spacingbetween the connections; and/or (c) inat-tention to the air space and creepage dis-tances needed when blocking and brac-ing are installed.

• The loose coil in the slot problemsmay be due to a slot content design thatdoes not take into account the gradual

shrinkage of insulating and wedgingmaterials, or where the need for tighcoils has been sacrificed to make the

coils easy to install in the slot.Probably the best way to avoid pre

mature stator winding insulation problems is to have an adequate purchasetechnical specification. Some suggestions for terms to include in such a purchase specification, in addition to the relevant parts of IEC 60034, are:

• For a 30-year life, require a Class Finsulation system to be operated at aClass B temperature rise.

• Require that the groundwall insulation system pass a voltage endurance

type test similar to those specified inIEEE Standards 1043 and IEEE 1553(regrettably there is no IEC equivalenspecification that is well defined and hasa pass-fail criteria). Requiring a voltageendurance test is probably better thanspecifying the maximum design electricstress, since this may retard the introduction of new materials and processes. Forfurther assurance, require spare coilsfrom the production batch for a stator tobe subjected to a voltage endurance test.

• Require a partial discharge test on

the new winding, together with a ‘blackout’ test to ensure the coils are properlyimpregnated, and clearances in the endwinding are sufficient [4,5].

• For multi-turn coils, require a volt-age surge test both on coils (for non global VPI stators) and complete windingsaccording to IEEE 522 (IEC 60034 Par15 is not as effective in detecting weakturn insulation).

• For non-global VPI stators, requirethe use of a wedging or sidepacking sys-tem that contains a follow-up restrain

Stator Windingcontinued from page 25

Figure 6: Photo of a coil removed from a hydrogener-ator stator that failed due to slot discharge. The verti-cal stripes are where the insulation is not abradedsince the stator core ventilation ducts are at these loca-tions.

Figure 5: Partial discharges occurring inthe inadequate space (here filled with aninsulating board) between two phases where the high voltage bars are connectedto the circuit ring busses. (Courtesy C.Maughan)


http://slidepdf.com/reader/full/issue106-example-of-premature-stator-winding-failure-in-recentrly-manufactured 4/4

that ensures tightness as the slot contentsshrink. This could include the use of twoor three part wedges, ripple springsand/or conformable materials such as sil-icon rubber. Alternatively consider

requiring a clearance between the side of the coil and the core to be no more than0.1 mm or so.

• Insist on the right to make unan-nounced factory inspections during man-ufacture of the stator.

Note that most of the above termsmay increase the cost of the stator wind-ing, but will probably result in a longerwinding life and less maintenance overthe lifetime. The machine owner also hasa responsibility to operate the machinewithin specification, keep the windings

clean and tight, and preferably visuallyinspect the winding before the end of thewarranty period. It would also be benefi-cial if manufacturers could educate userson the trade-offs of cost vs. life theymake when designing a new winding.

CONCLUSIONS.

1. Problems such as coil abrasion inthe slot, electric stress relief coatingdeterioration and partial discharges inthe endwinding have lead to failures in asshort as 5 years of operation. This anec-

dotal information is supported by the factthat PD for some manufacturers is high-er for recently made machines than forsimilar machines made over 10 yearsago.

2. To avoid premature failures, usersof modern air-cooled machines shouldensure they have a good purchase speci-fication and ensure the manufacturer hasan appropriate QA program in place.

REFERENCES

[1] G.C. Stone, V. Warren, “Effect of

Manufacturer, Winding Age and Insula-tion Type on Stator Winding PD Levels”,IEEE Electrical Insulation Magazine,Sept 2004, p13.

[2] C.V. Maughan, “Root Cause

Diagnostics of Generator Service Fail-ures”, Proc IEEE International Sympo-sium on Electrical Insulation, Sept 2004,p154.

[3] G.C. Stone et al, “Electrical Insu-lation for Rotating Machines”, Wiley-

IEEE Press, 2004[4] G.C. Stone, “A Suggested

Approach for Specifying PD Testing as aNew Winding Acceptance Test”, Proc.IEEE Electrical Insulation Conference,

Oct 2005, p159[5] Draft IEC Standard IEC 60034

Part 27, “Rotating Electrical Machines –Part 27 PD Measurements on the StatorWinding Insulation of Rotating Machin-ery”, Oct 2005.

27January 2006

Figure 7: Bar abrasion due to loose windings in the slot of aturbo generator that did not have sufficient sidepacking or aradial follow-up wedging system.

Figure 8: Example of loose winding failure in a 200 MVA global VPI generator stator.

Documents

Issue1_06 Example of Premature Stator Winding Failure in Recentrly Manufactured Motors and Generators