ger-3751a - understanding, diagnosing, and repairing leaks in

GE Power Systems

Understanding, Diagnosing,and Repairing Leaks inWater-Cooled GeneratorStator Windings

Joseph A. WordenJorge M. MundulasGE Power SystemsSchenectady, NY

GER-3751A

g

This article is for information purposes only. The operation of individual plants involves many factors beyond GE’s con-trol; it ultimately rests with the plant’s operator/owner. GE disclaims any responsibility for liability for damages of anytype, i.e. direct, consequential or special that may be alleged to have been incurred as result of applying this informationregardless of whether the claim is based on strict liability, breach of contract, breach of warranty, negligence, or otherwise.

Contents

Abstract . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1General Considerations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1

Leaks in Stator Hydraulic Components and Connections. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2Leaks in the Stator Bar . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3

Leaks in the Clip Window Braze . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3Leaks Through Porosity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3Leaks in the Clip-to-Strand Connection . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3Leaks Through Serrated Spacers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4

Leaks Initiated by Mechanical Causes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4Erosion in Hollow Conductors. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4Modern Winding Design . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5

Leak Descriptions and Definitions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5Stator Winding Leak Testing Methods . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6

On-Line Leak Testing Methods . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6Monitoring YTV System Vent . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7Stator Leak Monitoring System (SLMS) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7

Off-Line (Maintenance) Leak Testing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7Stator Preparation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8

Hydraulic Integrity Test (HIT) Skid . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8Vacuum Decay Testing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9Pressure Decay Testing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10Helium Tracer Gas Testing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11Stator Bar Capacitance Mapping Testing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12

Summary of Recommended Tests . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13Industry Experience . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13Stator Winding Leak Repair Methods . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15

Component Replacement . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15Braze/TIG Repair . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15Epoxy Injection Repair . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15

Long Term Leak Solutions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17Global Epoxy Injection . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17Full Stator Rewind . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17

Understanding, Diagnosing, and Repairing Leaks in Water-Cooled GeneratorStator Windings

GE Power Systems ■ GER-3751A ■ (08/01) i

Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18Questions & Answers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18Appendix . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 22

Crevice Corrosion Leak Testing Decision Tree . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 22References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23List of Figures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23List of Tables . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23


GE Power Systems ■ GER-3751A ■ (08/01) ii

AbstractIn May 1991, GE issued TIL-1098 which recom-mended routine maintenance testing of water-cooled stator windings for leaks. Leaks in thegenerator water-cooled stator winding canaffect machine availability and insulation life.The keys to minimizing the negative impact of aleak are early detection and timely repair.Routine leak testing of the stator winding is thebest way to determine the existence of leaks.

Leak tests, such as Vacuum and Pressure Decaytests, and use of the Stator Leak MonitoringSystem (SLMS), help detect larger leaks. Amore sensitive test, such as Helium Tracer GasTesting, is necessary to find smaller leaks. Ifleaks are located prior to insulation damage,they can be repaired and the unit returned toservice.

Stocking of stator winding and hydraulic com-ponents plays an important role in reducingrepair time for those cases where winding dam-age has occurred as a result of water leaks.

This paper provides information about thecauses of water leaks, recommended testingmethods and maintenance, and possible repairoptions.

IntroductionStator bar insulation failures that occur in serv-ice can cause extensive collateral damage to thegenerator stator core and field. This can extenda forced outage several months, dependingupon the extent of the damage. Insulation fail-ures during routine electrical testing can stillhave a great impact on the maintenance outagescope and duration. However, those repairs aretypically shorter in duration than with an in-service failure.

Wet insulation is the most common cause of fail-ure for water-cooled stator bars. The severe

impact of a stator insulation failure on genera-tor availability demonstrates the importance ofa thorough, routine leak testing and repair planthat includes stocking of long lead-time windingcomponents.

The first goal should be to detect and locateleaks within the winding before they causeirreparable damage to the stator bars.

The second, and equally important goal, is torepair those leaks with a process that will have aminimal effect on outage length while at thesame time will provide a high level of confi-dence for service life.

While TIL-1098, "Inspection of Generators withWater-Cooled Stator Windings" provides impor-tant guidance to further improve the generatorreliability, this paper:

■ Provides information about thedevelopment of water leaks

■ Describes the recommended testingmethods and maintenance practices

■ Suggests repair options

General ConsiderationsDuring operation, the stator winding is subject-ed to an environment of thermal shocks, cyclicduty, corrosion, mechanical vibration, and elec-tromagnetic stresses, where the potential fordamage and methods of leak detection, are verydifferent. Water-cooled stator windings havethousands of brazed joints that are potentialsources of leaks.

Windings and components are carefully testedfor leakage throughout the manufacturingprocess. However, leaks can develop after a peri-od of service. Your local GE representative canprovide a copy of related publications, supplyadditional information, and provide qualifiedtechnicians and specially designed equipment


GE Power Systems ■ GER-3751A ■ (08/01) 1

to perform the recommended tests and inspec-tions. In particular, GE publication GEK-103566, "Creating an Effective MaintenanceProgram," contains a listing of standard testsand other recommendations for generatormaintenance.

At the start of an outage and prior to degassingthe generator, the stator cooling water systemshould be removed from service. This is to min-imize the potential for water ingress from waterleaks into the stator.

Leaks in Stator Hydraulic Componentsand ConnectionsLeaks in the generator stator hydraulic systemmay originate at any of the following hydrauliccomponents:

■ Copper tubing, pipes, pipingconnections, elbows, fittings, sleevejoints, tube extensions, connectionsleeves

■ TeflonTM hoses and hose fittings

■ Liquid cooled series loop brazes

■ Connection rings

■ Inlet/outlet headers, nippleconnections

■ Liquid-cooled high voltage bushingsand connections

■ "P" bar liquid connections and hoses, ifapplicable

Figure 1 depicts a typical stator end winding con-figuration. Small leaks in these locations typi-cally will not result in winding damage so longas hydrogen gas pressure is maintained abovethe stator cooling water pressure. However, cap-illary action can cause water to leak from a bareven though gas pressure is maintained higherthan the cooling water pressure. Larger leakscan be detected by on-line testing methods andVacuum and Pressure Decay Tests, which arediscussed in more detail later.

Even extremely small water leaks can be detri-mental if allowed to persist. This is particularlythe case if the stator water system is left in opera-tion during outages when the generator isdegassed. Under these circumstances, the pres-sure differential provides the greatest drive forwater to be forced into the groundwall insulation.



Header

TeflonHose

HydraulicConnection Stator Bar

Bar Clip

Electric and HydraulicConnection

GE

R3

75

1-1

Figure 1. Typical stator end winding configuration

Leaks in the Stator Bar

Leaks in the Clip Window Braze

During the manufacturing process, a window inthe clip provides access for the mechanicalassembly of the clip to the strand package. Acover is brazed to the clip and strand packageduring the clip assembly. Though unlikely,braze joints for this cover could potentially be asource of leaks.

Leaks Through Porosity

One of two types of clips (leaf or bottle type) isbrazed to the ends of water-cooled stator bars.Modern leaf or bottle clips are machined fromsolid copper blocks, assuring excellent qualityand mechanical integrity of the copper clip.However, older machines were built with castclips. Some of the hydraulic plumbing compo-nents used on older units with bottle clips werealso cast. These cast components may developleaks in localized areas of high porosity.

Leaks in the Clip-to-Strand Connection

A clip-to-strand braze connection is madebetween the stator strands and the bar clip. Thisbraze connection is made and leak-tested dur-ing stator bar manufacturing.

A number of leaking clips have been analyzedfrom different generators. The leak paths ineach of the clips analyzed were similar. In eachcase, the cross-sectional size of the leak pathstayed relatively constant over its entire length.This indicates that the depth of the penetrationinto the copper and adjacent braze was indeedlimited. However, the corrosion mechanism wasable to continue by driving down the length ofthe copper strand. This, coupled with the selec-tive attack of the phosphorous-rich braze alloy,indicates that the corrosion reaction needsphosphorous to "fuel" the process. The evi-

dence discovered here revealed a leak processthat initiates in the braze alloy at the inner sur-face (a crevice corrosion mechanism). Underthe right conditions the leak can change to cor-rosive penetration of adjacent copper (phos-phoric acid attack), with a limited depth andcross section.

GE's analysis of the crevice corrosion phenome-non in clip-to-strand braze joints shows it to bea two stage process:

■ Phase 1. The first stage of the corro-sion process consists of corrosion ofthe braze alloy inside the clip at thestrand ends. Water works its way intoexisting small voids at the braze surface(due to the "spongy" nature of thebraze) and stagnates. If conditions areright (i.e. void size), crevice corrosionof the braze takes place attacking thephosphorus rich phase of the brazealloy. Initial void sizes increase asmaterial corrodes. The water chemistrychanges to phosphoric acid as a resultof crevice corrosion of the braze alloy.

The Phase I process is limited by thecrevice reaching critical volume, solutionchemistry and precipitating out phos-phate salt along the surface of the braze,which slows/stops further corrosion.

■ Phase 2. The second stage of corrosionbegins if the copper strand is contactedby the void as it grows. This causes acorrosive attack of the copper strandsby the phosphoric acid solutioncreated by corrosion of the braze alloy.The corrosion of the copper strandstakes place preferential to the brazealloy and at a higher rate. The depthinto the copper is limited and roughlyconstant (approximately 0.015 inches),



partly by the precipitation ofphosphate salt on the copper, andpartly by the volume of the liquidbecoming too large to maintain thecritical acid concentration.

The corrosion mechanism changes (dueto a change in water chemistry) and thephosphorus-rich phase of braze is againattacked. Then, as the solution chemistrychanges due to phosphorus being addedto the solution, preferential attack of thecopper occurs. The mechanism switchesback and forth from crevice corrosion ofthe braze alloy to acid attack of the cop-per as the leak path makes its way downthe strand. Figure 2 illustrates the crevicecorrosion process.

Leaks Through Serrated Spacers

Stator bar clips in the 1960s used a "serratedspacer clip" design. This design did not usefoam metal alloy filler and did not have highlevels of porosity in the clip to strand brazejoint. The initiation and development of crevicecorrosion in the "serrated spacer clip" designhas been a much slower process. These units

are starting to experience leaks from crevicecorrosion along the spacer clip braze joint afterthirty years of operation.

Leaks Initiated by Mechanical CausesThe end winding support system is designed tomove as one assembly. When signs of greasingor dusting are detected it is an indication thatthere is relative motion between components.Depending upon the location and severity ofobserved movement, corrective action mayneed to be taken immediately in order toreduce the risk of a high cycle fatigue failure ofthe copper plumbing. These types of leaks willresult in a noticeable change in the amount ofhydrogen detected in the YTV vent (the typicalname for the vent line to the atmosphere) fromthe stator cooling water storage tank. (See Figure3.) When this type of failure occurs on the inletside of the cooling system, cooling water can bedisplaced by hydrogen in the stator bar.Excessive bar temperatures may result leadingto insulation failure.

Erosion in Hollow ConductorsWater flow through the stator bar hollow con-ductors may result in some erosion of the cop-



1) Solution at braze surface (inlet water)2) Water works its way into voids in braze and stagnates3) Crevice corrosion of braze (primarily Cu3P)4) Critical solution or surface area conditions met, and Cu corrodes5) Solution/surface area conditions change favoring braze corrosion6) Solution/surface area conditions change favoring Cu corrosion7) Leak drives down the bar to outer surface

Voids in "Spongy"Area of Braze

Cooling Water

Copper Strand

Copper Strand

Brazing Metal

WA

TE

RB

OX

Selective Attackof Cu3 P

GE

R3751-2

Figure 2. Crevice corrosion mechanism

per. However, for GE generators operatingunder normal conditions, the rate of erosion isquite small, and will not impact operation overthe life of the winding. GE water-cooled gener-ators have not experienced a forced outage orrequired a stator rewind as a result of erosion.

Modern Winding DesignModern winding design provides resistance towater damage. Four design features minimizewater-leak related damage to the generatorwinding insulation during operation andbetween maintenance tests:

■ Inherent resistance of modern statorbar insulation systems to waterdamage. Stator bars insulated with

MicapalTM II exhibit excellenttoughness, durability, and resistant toexternally applied chemicals,moisture, oil, and other contaminants

■ Operation with gas pressure higherthan water pressure, which, in case ofa leak, results in hydrogen flowing intothe stator bar, and, from there, to theYTV vent.

■ Use of machined copper fittings.

■ Improvement in brazing processes.

Leak Descriptions and DefinitionsThere are a number of unknowns, variables,and assumptions that are made when discussing



COOLERS

GENERATORFRAME

WATERSTORAGE

TANK

LIQUIDDETECTOR

TURB.END

OUTLETHEADER

PUMPS

YCD

YST-1

YCFD YCFD

YCF

COLLEND

YTVVENT

INLETHEADER

GE

R3

75

1-3

Figure 3. Stator cooling water system

leaks and their effects. Development of leaks,rate of growth, and effects, are dependent uponvariables in operation, manufacture, and mate-rials. These variables make it difficult to deter-mine a "critical" leak size, water migration rates,and other parameters. When making these cal-culations, it is assumed that the ideal gas andfluid equations apply. An in-depth analysis ofthe crack propagation or fluid flow is beyondthe scope of this paper.

Leaks encountered in braze and weld joints typ-ically fall between 10-5 and 10-2 std. cc/sec.Leaks below 10-7 std. cc/sec will usually beplugged by the water itself. Bubble Testing andVacuum/Pressure Decay Testing are useful forleaks greater than 10-4 std. cc/sec; more sensi-tive detection tests (such as the Helium TracerGas Test) are needed for leaks smaller thanthat. These values apply for exposed leaks.Since the majority of leaks occur in the seriesloop connections, which are covered with manylayers of tape and epoxy, leak detectionbecomes more difficult as the sensitivities of alltests are reduced.

Stator Winding Leak Testing Methods

Stator leak testing can be broken into two dis-tinct categories: on-line testing and off-line (ormaintenance) testing.

On-line tests are defined as those tests that donot require removal of the generator from serv-ice. Off-line (Maintenance) tests define thebulk of the leak tests. These tests requireremoval of the generator from service and, insome cases, removal of the generator field.

On-line testing looks at the winding as a wholeand is not capable of isolating individual leaks.It is generally impossible to determine if the

detected "leak" is the result of one "large" leak,or several "smaller" ones.

Maintenance testing, on the other hand, has thecapability of locating individual leaks as small as10-5 std. cc/sec. Under the same conditions asthe "large" leak, this size leak would pass about10 std. cc/year. The size of a leak necessary tocause stator bar insulation failure is believed tobe on this order. Continued operation will onlyincrease the magnitude of the leak. The rate ofthis growth is a function of many variablesincluding location, material properties, andthermal and mechanical stresses.

Uncertainties over critical leak size and growthrates has prompted GE’s "early detection andrepair" philosophy.

On-Line Leak Testing MethodsOn-line testing allows for monitoring of thewinding condition over the period betweenmaintenance tests, but is not capable of isolat-ing individual leaks. However, on-line testing isstill an important part of proper stator mainte-nance. Early detection and repair are crucial tominimizing the damage that can result fromwater within the generator. Periodic monitoringof these indicators should be a fundamentalpart of operating all generators with water-cooled stator windings.

Even after the winding has passed routine main-tenance tests, large leaks caused by componentfailure or rapid propagation may occur duringoperation. Detecting and repairing these leaksquickly is important to minimizing generatordamage. Two fundamental on-line indicatorsare used to detect stator winding leaks:

■ The liquid detector alarm. Water foundin the alarm’s drain indicates thepresence of water within the generator.There may be several sources of this



water including the hydrogen seal oilsystem, the hydrogen coolers, statorwinding leaks, and even wet hydrogengas due to poor-quality of the gassupply.

■ The presence of hydrogen at the YTVvent. (See Figure 3.) If water is found inthe liquid detector, the YTV ventshould be checked for hydrogen. Thewater is likely to be coming from astator winding leak if the vent flow testresults indicate a problem. If the venthydrogen flow is "normal", an alternatesource of water should be explored.

Monitoring YTV System Vent

Excessive hydrogen gas flowing out of the statorwater cooling system vent indicates that a largeleak within the stator winding is likely. The sta-tor water cooling system is a hydraulically closedsystem except for the YTV vent, which isdesigned to allow for water expansion and con-traction caused by temperature changes duringoperation. The deionized water remains aerat-ed by virtue of the YTV vent. Since hydrogenpressure is normally higher than water pres-sure, a leak in the winding may show up asexcessive hydrogen flow through the vent.Hydrogen pressure inside the generator is theonly possible source of hydrogen to this system.

Accurate measurements of hydrogen flow arequite difficult and subject to error. Also, someuncertainty is present when attempting to dif-ferentiate between hydrogen flow due to a leakand that due to TeflonTM hose permeation.During operation, the hydraulic hoses can bethe source of up to 2 ft3 of hydrogen flow perday. Through routine monitoring of the ventflow, winding leaks can be detected by noting asteady upward trend or step increases in hydro-gen flow through the vent. GE recommends

checking the YTV vent on weekly basis. For ahydrogen flow rate of 10 ft3 to 200 ft3 per day,a full set of HIT Skid tests is recommended atthe next outage. If the flow rate is greater than200 ft3 per day, it is recommended that the unitbe removed from service and source of the leakrepaired.

Stator Leak Monitoring System (SLMS)

GE has developed a Stator Leak MonitoringSystem (SLMS) which is highly recommendedfor:

■ Oxygenating stator cooling water tothe recommended level

■ Monitoring the level of hydrogenescaping out of the YTV vent

The system consists of a flow meter, gas analyz-er, data acquisition and control system and a sys-tem piping modification package. The SLMSmodule is mounted at the hydrogen detrainingtank and connects to a gas analyzer and a flowmeter which are added to the existing piping.The system brings fresh filtered air into thecooling water to provide a measurable gas flowand to maintain the proper water oxygen con-tent (2 to 8 ppm). Figure 4 shows the typicalconfiguration of the SLMS system.

Measurement of the hydrogen content and gasflow provides an accurate measurement ofhydrogen leakage through the stator winding.The level of hydrogen leakage is directly relatedto a leak in the water-cooled stator winding.

Additionally, SLMS aids in minimizing statorbar copper erosion, resin bed damage, rectifiergrounds, and stator winding strand blockage.

Off-Line (Maintenance) Leak Testing A periodic off-line stator leak maintenance testprogram will indicate the presence of problemsand possibly avert failures altogether.



Maintenance testing of water-cooled generatorstators has been divided into two categories:

■ Tests performed during minor outages(typically every 30 months)

■ Tests performed during major outages(typically every 60 months)

The standard minor outage test program con-sists of a Vacuum Decay Test and a PressureDecay Test, in addition to visual inspection.These tests do not require access to the statorbore and can be performed with the rotor inplace, whenever the generator is removed fromservice.

The standard major outage test program con-sists of Pressure Decay Test, Vacuum Decay Test,Helium Tracer Gas Test and Stator BarCapacitance Mapping Testing, as well as electri-cal testing. During a major outage the field isremoved and this facilitates the performance ofa Helium Tracer Gas Testing and Stator BarCapacitance Mapping Testing. An outline formajor outage leak testing procedures is given inFigure 5.

Stator Preparation Having a dry stator hydraulic system (bars,headers, fittings, etc.) is essential to performingStator Leak Testing. The presence of moisturewithin the winding can conceal a small leak,making it undetectable to some leak tests. Themost efficient method of removing water is toperform a "Stator Blowdown" using very dry air.There are situations where the last remainingmoisture trace within the winding must beremoved by pulling a vacuum on the system,which will "boil off" the water. This is a time-consuming process and can be minimized byperforming a thorough blowdown prior to vac-uum drying. This process is slow in relation toblowdown, but is necessary. Typically, if a wind-ing has been dried well by blowdown, vacuummust be pulled on the winding until it is dry,which can take approximately 24 to 36 hours.

Hydraulic Integrity Test (HIT) Skid

To facilitate and expedite the dryout of thewater-cooled stator windings, as well as theVacuum and Pressure Decay Tests, GE has



PLANT COMPUTER

VENT GAS

SLMSSystem

WATERSTORAGE TANK

WATER COOLEDSTATOR

WATERRETURN

YTV

DRAIN

SCWS ALARM

POWERIN(110VAC)

Plant Air Supply6.5 CFM @ 100 psi PUMP

COOLINGWATER

AIR INJECTION

6 CFD

6 CFD

GE

R3

75

1-4

Figure 4. Typical Stator Leak Monitoring System (SLMS)

developed a self-contained, skid-mountedequipment, called a "HIT Skid" (HydraulicIntegrity Test Skid). Hoses from the skid areconnected to the generator at specified flangesof the stator. The only external requirement forthe skid is a 480 VAC power source at 30 amp.Figure 6 shows a typical Hydraulic Integrity TestSkid set.

Vacuum Decay Testing

Vacuum Decay Testing is a useful tool for deter-mining the integrity of the entire water-cooledstator hydraulic system. The primary advantageto vacuum decay testing is its sensitivity. Decaymeasurements are made in units of microns.(See Figure 7.) As a reference, one micron isequivalent to 0.00002 psi, which is undetectableon a typical pressure gage, yet easily measuredwith common vacuum gages. The test also meas-ures the leak rate of the entire winding withoutrequiring access to the stator winding.

Vacuum Decay Testing is relatively insensitive tochanges in temperature and atmospheric pres-

sure, and accurate results can be obtainedin one hour of decay testing. This is muchless than the typical twenty-four hours requiredfor a Pressure Decay Test. Ironically the test’ssensitivity can also be an obstacle to accuratetest results. Because the test is so sensitive,extremely small leaks at flanges and other non-welded connections, and the effects of out-gassing, can result in poor test results. For thisreason, it is very important for all connectionsto be tight and all components (hoses, connec-tions, flanges, gaskets, etc.) to be in good con-dition.

Moisture diffused into the copper oxide surfaceon the stator bars and the walls of the TeflonTM

hoses will outgas at a slow rate. This rate can behigh enough to cause the winding to fail thevacuum decay test. By analyzing the vacuumdecay data, it is possible to detect the pressurerise due to water outgassing. (See Figure 8.)Windings that fail Vacuum Decay Testing andshow indications of outgassing must be furthervacuum-dried and retested.



Stator Preparation

FailedCapTest

PassedCapTest

Visually Check forMoist HydraulicCircuit,

Series Loops,or Leaks in LocationsNot Checked w/Tracer

No Risk of In-Service or

Electrical TestFailure Due

to Wet StatorWinding

CarefullyInspect Suspect

Series Loopand Winding

UsingTracer Gas

VisuallyInspect Series

Loops andWinding for

MoisturePerform

CapacitanceMapping

Locate Leaks w/TracerSniff or Snoop

Tracer Gas TestTracer Gas Test

Capacitance Mapping Test

Repair Leaks

Retest withTracer Gas

Retest Vac & PressDecay (or HITS)

Vac & Press DecayTests (or HITS)

ELECTRICALTESTING

PASS

PASS

PASS

PASS

FAIL

FAIL

FAIL

FAIL

GE

R37

51-5

Figure 5. Major outage leak test plan

Pressure Decay Testing

The Pressure Decay Test has two advantages incomparison to the Vacuum Decay Test. It pro-vides a greater pressure differential (at least fivetimes that of the vacuum test) and applies pres-sure in the normal direction of the leak flow.These factors may make it easier to find leaksundetectable to vacuum. During the PressureDecay Test, exposed potential leak sites can betested with a "bubble" solution.

Drawbacks to pressure testing are its insensitivi-ty to small leaks, sensitivity to changes in theenvironment (temperature and barometricpressure), and the time required to test. In atypical test at 80 psi, a volume of 1.0 ft3 mustleak out of the generator to register a change of1 psi. For this reason, extremely accurate gagesare needed, preferably with 0.1 psi indications.Also, to detect small leaks, the test must be doneover many hours so that the leak volumebecomes significant relative to the test’s sensi-tivity.

Prior to exposure to high pressures, the statorwinding must be completely dry internally.Subjecting a winding with water in it to highpressure may force moisture into insulation,through yet undetected leak paths, and cause



Figure 6. HIT Skid

10–9

10–8

10–7

10–6

10–5

10–4

10–3

10–2

0.1

1

2

2

2

2

2

2

2

2

2

5

5

5

5

5

5

5

5

5

10–8

10–7

10–6

10–5

10–4

10–3

10–2

0.1

1

2

2

2

2

2

2

2

2

5

5

5

5

5

5

5

5

10–6

10–5

10–4

10–3

0.1

1

2

2

2

2

2

5

5

5

10–2

2

5

10

10

2

3

2

2

10

2

5

10

2

5

10–7

10–6

10–5

10–4

10–3

10–2

0.1

1

2

2

2

2

2

2

2

5

5

5

5

5

5

5

10

2

2

5

5

5

5

5

INC

HE

SO

FM

ER

CU

RY

PS

IA

TO

RR

AT

MO

SP

HE

RE

GER3751-7

Figure 7. Units of pressure conversion nomograph

GER3

751-

6

insulation damage. Small leaks can also be con-cealed by moisture within the leak path.Therefore, the winding should be initially vacu-um-decay tested in order to assure dryness. Thepressure test should be conducted with very dryair or nitrogen to assure that moisture is notreintroduced to the system.

Helium Tracer Gas Testing

Helium Tracer Gas Testing is a method of leakdetection where the generator is pressurizedwith a helium gas so that possible leak pointscan be detected using a helium gas detector.There is a wide range of tracer gases and tracergas detectors on the market.

The Mass Spectrometry technology used by GEemploys helium as the tracer gas because it isthe lightest inert gas, nontoxic, and non-haz-ardous. Other gases do not provide the level ofsensitivity of helium, and some of them cancombine with any residual water in the windingto form acidic solutions.

Tracer gas detector sensitivity is very importantin finding leaks in the generator stator winding.Leak sources can be buried beneath several lay-

ers of glass, mica, and resin within the winding.This can make detection difficult. A process ofbagging the series loop connections (the sourceof most leaks) has greatly improved the abilityto locate very small leaks. Helium pressure ismaintained on the winding for a period of timeto allow helium from a buried leak to migratethrough the insulation and become concentrat-ed in the bag. In about 85% of the cases, afterremoving the insulation, leaks on the order ofat least 10-4 std. cc/sec have been found. Theother 15% generally showed traces of helium invarious locations not concentrated enough topinpoint for repair.

In many cases, leaks that were found with thetracer gas would have been missed if the statorwinding only had been Vacuum Decay Testedand Pressure Decay Tested. Early detection pro-vides the opportunity to make repairs beforemore extensive damage can occur to the statorwinding insulation.

Tracer gas testing does have its own shortfalls,however, which highlight the importance of aprogram inclusive of Vacuum Decay Testingand Pressure Decay Testing. The two most



Outgassing/Virtual LeakSource is Depleted; RealLeak Takes Over

PR

ES

SU

RE

TIME

GER3751-8

Figure 8. Rate of rise curve (outgassing and leak)

notable shortfalls are the inability to inspect theentire winding, and the need for access to thewhole winding. To detect small leaks, the snifferdetector must be brought within 2 to 3 inches ofthe leak. Since it is nearly impossible to coverevery square inch of the winding, tracer testtechniques such as bagging the series loops, testonly the most probable leak sites. This cannotprovide confidence that the entire winding isleak-tight. Under normal circumstances it ispreferable to remove the field to performHelium Tracer Gas Testing. Limited testing canbe done with the field in place with the upperend shields removed.

Stator Bar Capacitance Mapping Testing

Stator Bar Capacitance Mapping Testing inwater-cooled generators is recommended todetermine the extent of water penetration ofthe groundwall insulation. The stator windingdoes not have to be dry to perform theCapacitance Mapping Test. This test is non-destructive to the stator bar. If a bar fails theCapacitance Mapping Test, it should be consid-ered unsuitable for long-term service. Theintent of this test is to locate bars that are athigh risk of in-service and/or hipot failure.

Capacitance Mapping Test results are sensitiveto the number of layers of bar insulation. Thissensitivity makes it necessary to test the bars ina location where the groundwall insulation isconsistent from bar to bar, typically at the "firstbend." The series loop joints may not be ade-quately tested with a capacitance probe becauseof the possible inconsistencies in the taped insu-lation thickness. This is because series loopsmay be unevenly hand-taped while the statorbars are evenly machine-taped. Design of thestator end winding makes access to the statorbar insulation difficult except at the first bendregion, which is near the stator core.

All insulating materials (i.e., air, MicapalTM,water) have a property known as the dielectricconstant. The Capacitance Mapping Test isbased on the difference between the dielectricconstant of water and that of the groundwallinsulation system. Where there is a mixture ofair, water, and insulation, the measured capaci-tance will be a different composite number. TheCapacitance Mapping Test uses this principle toidentify insulation that has been penetrated bywater. The capacitance probe is attached to thebars within three inches of the bar armor.Capacitance is measured between ground andthe test area. Before the test, the bars must becleaned with a solvent to assure proper contact.Although Capacitance Mapping Testing onlyinvestigates the condition of the bar insulationimmediately under the probe, if a bar is leaking,it is anticipated that the water will migrate alongthe bar to where the electrode is located.

Capacitance readings are taken for each of thebars in the winding, and at both ends of thegenerator. Good capacitance data will providea normal curve when plotted with nearly all ofthe data falling between –2 and +2 standarddeviations from the average. Wet bars that havefailed electrical testing have generally hadcapacitance values falling outside +5 standarddeviations from the average. Industry feedbacksuggests that bars found with a capacitance levelgreater than +3 standard deviations are to beconsidered suspect.

Bars that are found with damaged groundwallinsulation are not repairable and requirereplacement within one year because long-termexposure to moisture weakens the insulation,making it more susceptible to mechanical dam-age and failure from normal and transient volt-age stresses. Also, moisture within a strand pack-age deteriorates the bonding resin betweenstrands and may lead to separation of packages.



Relative motion between strands could nowcause strand shorts and ultimately bar failure.There is no effective way to dry a wet bar.Damage is irreversible.

Summary of Recommended Tests

Table 1 summarizes the recommended leak testsfor water-cooled generator stator windings, aswell as their frequency of performance. Therecommended outage intervals have proventhemselves over many years of use.

Industry ExperienceA review of historical data compiled on water-cooled generators through May, 2001, identi-

fied stator winding leaks as one important fac-tor affecting generator reliability and availabili-ty. Through May, 2001, there were approxi-mately 600 GE water-cooled generators in serv-ice, with an average stator winding age of over26 years. The expected design life of a liquid-cooled winding is 30 years.

Through May, 2001, of the number of water-cooled stator windings that failed, 15% failed inservice, and 55% failed electrical tests. The in-service failure rate has shown a marked increaseover the past several years. This appears to bedue to changes in maintenance practices.

Figure 9 shows a breakdown by cause of failurefor water-cooled stator winding failures throughMay, 2001.



Recommended Testing Frequency

On-Line During Outages

TestWeekly (End Shields (with Rotor

Removed) Removed)

Minor Outage Major Outage

LiquidDetector X

YTV Flow X(2)

VacuumDecay Test X(1) X

PressureDecay Test X(1) X

Helium Test

Helium Test withRotor Removed X

CapacitanceTest X

NOTES:

(1) For generators not equipped with SLMS. Vacuum and Pressure Decaytests are recommended during minor outages to assess the integrity ofthe stator hydraulic system.

(2) For units equipped with SLMS, the YTV vent is monitoredautomatically.

Table 1. Recommended leakage tests for water-cooled generator windings

A review of the leak-testing history provides evi-dence that passing both the Vacuum Decay Testand the Pressure Decay Test—though vital inthe Maintenance and Inspection Program ofWater-Cooled Generators—does not necessarilyguarantee a leak-free stator winding. The resultsdemonstrate the importance of Helium Tracer

Gas Testing and Capacitance Mapping Testing.(See Figure 10.) The combination of all thesetests provides the best possible assurance thatthe stator windings are free of destructive waterleaks. Our experience indicates that of thosegenerators with windings that failed electricaltests due to wet insulation through May, 2001:



Electrical Test

In Service

7

6

5

4

3

2

1

0

Nu

mb

er

of

Even

ts

1985 1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001

Comments for In-ServiceFailures:

1989 One failure probably due towater leak. Causeunknown.

1995 Two failures due toabrasion.

1999 One failure due to crevicecorrosion. One failure dueto abrasion.

2000 One failure due to leakingcopper tube and/or waterconductivity.

2001 One failure due to crackedtube.

Water-Cooled Stator Bar Failures

75% of Failures Due to Wet Insulation

Figure 9. Updated leak test data

PASSED

FAILED

Vacuu

m

Pre

ssur

e

Trac

er

Cap

acita

nce

100%

90%

80%

70%

60%

50%

40%

30%

20%

10%

0%

82%

18%

37%

67%

30%

63%

33%

70%

GE

R3

75

1-1

0

Figure 10. Maintenance test data through May, 2001

■ Most passed the Vacuum Decay Test

■ Approximately 50% passed the PressureDecay Test

■ All failed the Helium Tracer Gas Testand Capacitance Mapping Test

Stator Winding Leak Repair MethodsStator winding leak repair methods can be bro-ken down into three categories:

■ Component Replacement

■ Braze/TIG Repair

■ Epoxy Injection Repair

The type of repair method depends upon anumber of factors including type of leak, itslocation, leak size, the desired longevity andconfidence of the repair, and the condition ofthe bar insulation.

Component Replacement

Components that have required replacement inthe past include TeflonTM hoses, hydraulic fit-tings and piping, and stator bars.

Leaks in hoses, fittings and piping connections,are usually the result of fatigue failure, partsabrasion, or casting porosity. These compo-nents can frequently be replaced with the gen-erator field in place. However, there are somewinding arrangements that limit access and dorequire field removal from the stator to removeand install new parts.

Stator bar replacement is a more complex task,and does require the field to be out of the sta-tor. For that reason bars are replaced only whenthe fundamental integrity of the groundwallinsulation has been compromised.

Most water leaks have been found and repairedprior to damaging the electrical insulationproperties. However, when a leak occurs at the

clip-to-strand braze, water can migrate alongthe bar at the interface between the bar copperand insulation. This can result in water accu-mulation in the insulation, particularly at thecore line, and result in an electrical failure toground during operation or failure duringmaintenance testing. Insulation degraded bywater permeation cannot be restored to origi-nal condition, and replacement of the bar isrecommended if the "wet" insulation is detect-ed in the first bend near the core line in the barslot or grading section.

As demonstrated by industry experience, a stock-ing program of stator bars and long-self life mate-rials reduces outage time extension in caseswhere major repairs need to be performed dueto failed bars and water leaks. Table 2 shows therecommended stocking quantities for spare sta-tor bars.

Braze/TIG RepairLeaking hydraulic braze joints are usually fixedby rebrazing, and additional braze alloy isadded, as required. Tungsten Inert Gas (TIG)repairs are used on porosity leaks and repairingwindow leaks. TIG repairs are done on windowleaks because the stator bar insulation may bedamaged if heated too much.

The TIG repair causes only localized heatingand should not damage bar insulation.However, TIG repairs are largely surfacerepairs, making them mostly suitable for loca-tions where strength is not a requirement.

Epoxy Injection RepairClip to strand leaks can be repaired by GE’spatented Epoxy Injection Repair process. Thisprocess uses a specialized borescope with videodisplay in conjunction with specialized epoxyinjection equipment.



Figure 11 shows a typical epoxy-injected clip.The epoxy itself is a specifically formulatedcompound designed to bond to the oxide layeron the surface of the copper. The stator wind-ing must be dry prior to starting the epoxyinjection process. Access is gained by removingthe TeflonTM hose and the hydraulic pumpingcomponents. The technician directs a syringe’spinpoint nozzle to the target area aided by aborescope video display. The epoxy is injectedinto the crevices between the strands. Each dia-mond shape area formed by the strand corners,as well as between strand columns and rows, isinjected to fill any voids. Clips on both ends of

the stator bar need to be injected to ensure thatthe leak source is sealed. The epoxy is heat-cured and then the clip is inspected and leak-tested. Once the epoxy is cured and the unit haspassed the required testing, the procedure isconsidered to be a permanent repair. Thisrepair is performed by a qualified GE specialist.

Prior to performing the epoxy injection repair,the stator bar-to-bar insulation or bar-to-con-nection ring insulation is removed to verify thatthe leak is a clip-to-strand leak and not a clipporosity or window leak, which are repairedusing other methods. If a clip-to-strand leak isverified, it should be determined that the stator



Recommended Number ofSpare Generator Stator Bars

Four-Pole Two-PoleNumber of Slots Number of Slots

54 60 66 72 30 36 42 48 54 60

Top 14 15 17 18 15 18 21 23 25 27

Bottom 4 4 4 4 4 4 4 4 4 4

P-Bar* 1 1 1 1 2 2 2 2 2 2

StatorBar

*Machines with Generrex Excitation Only

Table 2. Recommended number of spare stator bars

Figure 11. Epoxy-injected clip

bar insulation is not wet. The CapacitanceMapping Test is a reliable indicator of wetinsulation. Insulation is electrically andmechanically weakened by water penetration.Replacement of stator bars having wet insula-tion is recommended due to the potential foran in-service failure and significant damage tothe generator.

In addition to the Capacitance Mapping Test,GE recommends that the suitability for serviceof the stator bars be confirmed with a highpotential test prior to performing this repair.The repair should only be performed on barsthat are considered suitable for service.

GE’s proprietary Epoxy Injection Repair is thesimplest, most non-invasive and highest tech-nology repair available. The procedure can beperformed during either a planned or emer-gency outage. To minimize the impact on out-age duration, the epoxy injection repair is timeefficient, flexible to respond to sudden changesin scope, low risk, and provides a high quality,reliable, and permanent solution.

GE will work with customers to develop a sched-ule that fits the specific outage window. Toobtain a summary of experience with thisrepair, contact your local GE office.

Long Term Leak Solutions

In the competitive environment that utilitiesface today, reduction in generator availabilityfrom unplanned outages and added mainte-nance (due to water leaks) may pose an unac-ceptable penalty. GE offers two options for along-term solution to address these leaks andprovide improved generator availability:

■ Global Epoxy Injection

■ Full Stator Rewind

Global Epoxy InjectionGlobal epoxy injection consists of the applica-tion of epoxy to all stator bar clips. This can beperformed on a planned basis, at one time, orover a series of outages. This proactive applica-tion to all the stator bar clips eliminates futurerisks of stator bar leakage due to crevice corro-sion, reduces the potential for outage extensionand the likelihood of in-service failure due towater leaks.

GE’s research indicates that—if unabated—crevice corrosion in the stator cooling systemwill continue to progress through the remain-ing life of the machine. On machines builtbefore 1986, the risk associated with the poten-tial development of leaks as the generators ageis considerable. The Global Epoxy Injectionprocess eliminates this risk and the need forcontingencies for leak repair and possible barreplacement. The cost of replacing a wet barcan be significant, considering replacementparts, labor, and extension of outages. For cus-tomers that do not wish to rewind the stator dueto crevice corrosion problems, we strongly rec-ommend global application of epoxy to all clipsas soon as possible.

Full Stator RewindWhile Global Epoxy Injection can eliminatefuture risk of stator bar leakage due to crevicecorrosion for a generator stator winding thathas not yet experienced a bar leak, a completegenerator stator rewind provides the best long-term solution to leaks in a water-cooled genera-tor. Thanks to the introduction of modernmaterials and to refinements and optimizationin design practices and technology, the reliabil-ity of a water-cooled generator stator windingcan be greatly increased. A full stator rewindalso provides the opportunity for an electricalpower uprate of the water-cooled generator.



ConclusionAs the typical water-cooled generator in theindustry approaches 30 years of service, there isan increasing likelihood of winding failuresassociated with water leaks. The cost can be sig-nificant in terms of added downtime for repairof generator stator cooling water leaks. Earlydetection and repair of water-cooled statorwinding leaks can significantly reduce the riskof winding insulation failure, as well as the costof downtime and winding repair. A proactiveapproach to on-line monitoring and off-lineleak testing is necessary for early detection ofleaks in order to maximize the service life of thestator windings.

Questions & AnswersThe following are typical questions and answersregarding generator stator cooling water leaks,testing, and repair. They are included to sup-plement the information provided in the mainbody of this paper.

1) Q. What are the causes of leaks in a water-cooled stator winding?

A. There are a number of possible / prob-able causes for the development ofleaks in a liquid-cooled stator winding.During normal operation the statorwinding is subjected to thermal shocks,cyclic operation, corrosion, andmechanical/electrical vibration. Thesenormal operational stresses can lead tothe development of leaks in a liquid-cooled stator winding.

2) Q. If we implement Global EpoxyInjection on a winding that has not yetseen crevice corrosion, will we beexempted from the periodic testingrequired by TIL-1098?

A. Implementing Global Epoxy Injectionwill eliminate any future risk associatedwith crevice corrosion/clip to standleaks. However, during normal opera-tion the stator winding is subjected tothermal shocks, cyclic operation, corro-sion, and mechanical/electrical vibra-tion. These normal operational stressescan lead to the development of otherkinds of leaks. Therefore, the testing inTIL-1098 is still recommended.

3) Q. What improvements have been madeby GE to reduce the likelihood of devel-oping leaks in water-cooled generatorstator windings?

A. GE has made continuous improve-ments that resulted in liquid-cooled sta-tor windings of high quality that are farmore reliable and more efficient thanever before. By applying our state-of-the-art processes, and by using superiormachined stator bar clips, advancedbrazing techniques, and enhancedcleaning methods, we are confident ofhaving a product that, when operatedand maintained as recommended, willprovide our customers with generatorwater-cooled windings with the highestreliability in the industry.

4) Q. We have just completed implementinga Global Epoxy Injection on one of our1970s vintage generators. What is theprobability of developing leaks onother sister units?

A. Liquid-cooled generators that weremanufactured between 1970 and 1986are most susceptible to developing clip-to-strand leaks as a result of crevice cor-rosion. Clip-to-strand braze connec-tions on liquid-cooled stator bars that



were made prior to 1970 contained ser-rated spacers and were much moreresistant to developing these clip-to-strand leaks. Now, however, after morethan 30 years of service, we have docu-mented cases of clip-to-strand leaks, theresult of crevice corrosion, in a few ofthese machines. The CreviceCorrosion Leak Testing Decision Treemay be used as an aid to planning gen-erator maintenance activities. (SeeAppendix.)

5) Q. Why is Helium Tracer Gas Testing rec-ommended following acceptableVacuum Decay Test and Pressure DecayTest results?

A. The Helium Tracer Gas Test is the mostsensitive leak test that GE offers to date.It is capable of detecting leaks nearlyfive orders of magnitude smaller thanthat which can be detected with eitherthe Vacuum Decay Test or PressureDecay Test.

6) Q. What size leak does GE consider accept-able?

A. Any leak that is detected using methodsdescribed in this paper are consideredunacceptable.

7) Q. Do the leak test criteria in TIL-1098apply to oil-cooled stator windings?

A. Our experience shows that oil-cooledstator windings have not been suscepti-ble to stator bar insulation failurecaused by stator winding leaks. For thisreason, oil-cooled machines were notincluded in TIL-1098. However, theleak testing methods recommended forwater-cooled machines could beapplied to oil-cooled machines, but athorough flush of the windings would

be required prior to testing. The otheroption is to continue with the testingrecommended in the instruction book.This has proven successful to date.

8) Q. Why does GE recommend removingwater pressure when the machine gaspressure is removed?

A. The winding water pressure shouldalways be maintained lower than themachine gas pressure to minimizewater penetration through leak pathsin the stator winding. This practice isprecautionary and makes sense, evenfor windings that have been successful-ly leak tested.

9) Q. Is it possible for a winding to pass someleak tests and to fail others?

A. Yes. Each of the recommended leaktests has its unique advantages and lim-itations. It is recommended that all testsbe conducted at the suggested intervalbecause of the complementary natureof these tests. Some leak types and/orlocations are more likely to fail certainleak tests. Therefore, it is recommend-ed that all tests be performed at thesuggested interval.

10) Q. Does the winding need to be dry toperform the Capacitance MappingTest?

A. No. The Capacitance Mapping Testdetects the presence of water within theground wall insulation. This test iseffective because of the large differ-ence in the dielectric constant of waterand the normal ground wall insulation.

11) Q. Why is the acceptable leak criterion dif-ferent for the Vacuum Decay Test andthe Pressure Decay Test?



A. The sensitivity of the Vacuum DecayTest, and the effects of winding mois-ture and test equipment condition ontest results have made it necessary toestablish a higher acceptable decayrate.

12) Q. Can water escape through a leak pathwith the machine under gas pressure?

A. Yes. Through capillary action water canescape through some leaks. Leaks thatare buried beneath layers of insulationand resin are believed to be more sus-ceptible to this phenomenon. Thehydrogen pressure drop across theselayers of tape is also believed to be a fac-tor.

13) Q. Why is bar replacement recommendedfor wet bars that pass hipot?

A. Damage to wet insulation is irre-versible, and there is no known way todry a bar once it has become wet. Longterm exposure to water weakens theinsulation, making it more susceptibleto mechanical damage and failure fromvoltage stress. In addition, the watercan cause the bonding resin that holdsthe copper strand package together todeteriorate allowing the strands to sep-arate and relative motion between thestrands to occur. This relative motioncan lead to strand shorts, causing over-heating and ultimately bar failure.

14) Q. If Tracer Gas testing is the most sensi-tive test available, why are VacuumDecay and Pressure Decay Testing nec-essary?

A. There are several reasons why VacuumDecay and Pressure Decay tests are nec-essary: (a) First, tracer gas testingrequires access to the stator winding,

which is generally available only atmajor outages. Vacuum Decay andPressure Decay Testing do not requireaccess and can be performed at anytime with the generator off-line, pro-viding the ability to test the windingbetween major outages. (b) Tracer gastesting is used to inspect easily accessi-ble locations (such as the series loopand header connections) whichaccount for the majority of windingleaks. Vacuum Decay and PressureDecay procedures test the entire wind-ing, and are needed to test locationsnot covered by tracer testing. (c) Littleadditional effort is needed to performVacuum Decay and Pressure Decay testswhen preparing to tracer gas test awinding. Since the winding must bevacuum dried prior to tracer gas test-ing, Vacuum Decay Testing does notrequire additional equipment. PressureDecay Testing can be performed whilethe winding is pressurized with tracergas, thereby adding little time to theoverall inspection. (d) The combina-tion of these three test methods withthe Capacitance Mapping Test, pro-vides the best possible assurance thatthe stator windings are free of destruc-tive water leaks.

15) Q. Does the stator winding have to becompletely dry prior to leak testing?

A. Yes. Research and experience hasshown that smaller leaks can beblocked by moisture at typical test pres-sures. Vacuum drying is necessary tocompletely remove all moisture withinthe winding prior to testing.



16) Q. If we install SLMS will we be exemptedfrom performing the periodic testingrequired by TIL-1098?

A. No. The Stator Leak MonitoringSystem (SLMS) provides continuouson-line monitoring for leaks, but itdoesn’t have the same level of sensitivi-ty that can be achieved by the tests rec-ommended in TIL-1098. Because of thecomplementary nature of these tests, itis recommended that these tests be per-formed at the suggested intervals.

17) Q. If we rewind our generator stator usingmodern materials and techniques, orinstall a whole new generator, will we beexempted from performing the period-ic testing required by TIL-1098?

A. Even though windings and compo-nents are carefully tested for leakagethroughout the manufacturing process,those tests cannot assure that waterleaks will not develop after a period ofservice. During operation, the statorwinding is subjected to an environmentof thermal shocks, cyclic duty, corro-sion, mechanical vibration, and electro-magnetic stresses. Slight variations incomponent and braze qualities can alsoresult in water leaks after exposure toan operating environment. It is recom-mended that all of the tests recom-mended in TIL-1098 be performed, atintervals suggested by experience,because of the complementary nature

of those tests. Some leak types and/orlocations are more likely to fail certainleak tests. In order to test for all leaktypes and locations, all the leak testsreferenced in this paper are recom-mended.

18) Q. I have heard that low level oxygen sta-tor cooling water systems do not havecrevice corrosion. Does GE recom-mend converting its high oxygen levelsystem to a low oxygen level system?

A. Some studies indicate that low-leveloxygen systems are less susceptible todeveloping corrosion. However, thosesystems have problems. Low-level oxy-gen systems are more susceptible todeveloping cuprous oxide. This is anunstable oxide that tends to plug hol-low conductors in stator bar, blockingthe cooling water flow. These pluggedstrands can lead to overheating of thebar, separation and delamination of thestrand package, which may lead to abreakdown of the groundwall insula-tion, resulting in a stator ground fail-ure. GE does not recommend convert-ing from its high-level oxygen system toa low-level oxygen system.

19) Q. What is the recommended concentra-tion of dissolved oxygen in the statorcooling water?

A. The recommended concentration fordissolved oxygen in the stator coolingwater is 2 to 8 ppm.



Appendix



Action Risk Next Step

NextOutage

RewindLowest

PossibleRisk

• Fixes leak issues• Potential output/efficiency improvement• Incorporates state-of-the-art technology

Results/Benefits:

Outcome/Future Action

Next Outage,Repeat Review

of All Four Options

See Below forResults/Benefits

Reduced Testingwith SLMS

Repeat Every Outage.Ultimately, Rewind

or Repair

• Reduces risk of outage extension• Reduces followup testing w/SLMS

Test

PR

E-O

UT

AG

EM

AIN

TE

NA

NC

EP

LA

NN

ING

HighRisk

Ultimately, Bar Failures

See Belowfor Results of

Test or Rewind

Nothing NextOutage

Nothing

Rewind

Test

Epoxy InjectionLeak Repair

Wet BarReplacement

Full Rewind

Global EpoxyInjection

Med Risk

Med Risk

No Risk

Low Risk

Nothing

Test

Test

Rewind

Global EpoxyInjection

LowRisk Permanent Leak Repair If No Wet Bars

Leak– High –

Risk

No Leak– Med –

Risk

CREVICE CORROSION LEAK TESTING DECISION TREE

References1. Anderson, J. Armature Bar Insulation Capacitance Measurement Procedure. General Electric

Co., July 1990, p.2.

2. Lembke, B. Liquid Leak Test and Repair. General Electric Co., July 1990, pp. 12-18.

3. Varian Associates, Incorporated. Introduction to Helium Mass Spectrometer Leak Detection.Palo Alto, CA: Varian Associates, 1980.

4. Zawoysky, R. Stator Water Cooling System Issues. General Electric Co., January 2001.

5. Worden, J. Updated Leak Test Data. General Electric Co., May, 2001.

6. TIL-1098, Inspection of Generators With Water-Cooled Stator Windings.

7. GEK-103566, Creating An Effective Maintenance Program.

List of FiguresFigure 1. Typical stator end winding configuration

Figure 2. Crevice corrosion mechanism

Figure 3. Stator cooling water system

Figure 4. Typical Stator Leak Monitoring System (SLMS)

Figure 5. Major outage leak test plan

Figure 6. HIT Skid

Figure 7. Units of pressure conversion nomograph

Figure 8. Rate of rise curve (outgassing and leak)

Figure 9. Updated leak test data

Figure 10. Maintenance test data through May, 2001

Figure 11. Epoxy-injected clip

List of TablesTable 1. Recommended leakage tests for water-cooled generator windings

Table 2. Recommended number of spare stator bars



ger-3751a - understanding, diagnosing, and repairing leaks in

Documents