THE UNCERTAINTY IN POWER TRANSFORMER FAULT DIAGNOSTICS USINGCONVENTIONAL TESTING METHODS

Yogendra Dev Vashishtha*, Paul Ascione* and Qi Su**

* Power and Water Authority

** Monash University

Abstract

The average age of power transformers in Australia is about 25-35 years. Conventional oil tests,dissolved gas analysis (DGA) of oil, dielectric loss angle (DLA or tanδ) of winding and recoveryvoltage monitoring (RVM) are among many diagnostic techniques available. Power and WaterAuthority (PAWA) of Northern Territory refurbished five 66 / 11 kV, 10—30 MVA, 24—33 yearsold transformers. From the test results it is observed that a transformer which has very goodresults from these diagnostic tests and even has minimum moisture content in the winding(from RVM) may still have very poor mechanical integrity. Winding cellulose paper may beaged to such an extent (assessed from degree of polymerisation; DP) that any sudden force ortransient would cause failure. DP test is very difficult to perform regularly to ascertainmechanical integrity of winding paper. Furan analysis is another alternative to DP. Most of thetimes Furans are not detected and there is no universal correlation available for DP and Furans.Also, both ageing and moisture content influence Furans. Frequency response analysis (FRA)gives information regarding winding movement but it does not give much information aboutmechanical integrity of a winding which is essential for transformer reliability. This paper alsocompares moisture contents in the winding assessed from RVM and estimated from oil PPM(using Norris diagram) and estimated from winding DLA (based on Gussenbauer’s relationship).It appears that moisture content from RVM needs further validation and research toseparate out ageing and moisture effects.

1. INTRODUCTION

Transformers represent a high capital investment in atransmission substation at the same time as being akey element determining the loading capability of thestation within the network. With appropriatemaintenance, including insulation reconditioning atthe appropriate time, The technical life of atransformer can be in excess of 60 years. The end oflife, however, can be strategic or economic.Quantitative, risk based approach can be used to aidcostly investment decisions involving transformer life,otherwise made from a subjective viewpoint.Diagnostic systems have been listed under threecategories as portable / periodic, continuous on-lineand specialist support [1]. There are various on-linemonitors for load, operating conditions, gas-in-oil andmoisture-in-oil are available these days [5]. Dissolvedgas analysis (DGA) is the main condition monitoringtechnique used at Power and Water Authority(PAWA) in the Northern Territory. Recovery VoltageMonitoring (RVM) has been found as a bettertechnique to estimate moisture content in the windingpaper insulation and its ageing. However, RVM doesnot distinguish between moisture content anddegradation of paper. Some experiments have beenreported correlating charging current with paperageing. Due to very limited work, these are not very

conclusive. There is a possibility that charging currentmay also correlate with insulation resistance (IR) aswe found RVM following polarisation index (PI).

2. BACKGROUND

2.1 The major factors involved in transformerinsulation life are:

• Insulation moisture content

• Extent of oil oxidation

• Extent of cellulose degradation—water, CO,

CO2 and furans

• Insulation Structure compression

MOISTURE, OXYGEN and HEAT are the threeageing accelerators.

2.2 Transformer Life Assessment Indicators:

Insulation moisture content in oil provides a simplemethod of estimating residual life. The followingguide may be useful (based on equilibrium watercontent of paper as a function of water content ofimpregnating transformer oil):

3 % Maximum acceptable

4 % Entering “ at risk “ zone

5 %--6 % Considerable risk of failure

7 % Failure imminent

For life assessment, emission associated with theinsulation ageing process is of interest, principally thelevels of CO, CO2 and furans in the oil. Cellulosedegradation can be assessed by furan (HPLC) analysisand preferably supported by degree of polymerisation(DP) for a more definitive assessment. DP testinvolves the removal of samples of paper from thewinding insulation. Due to the need for internal accessand expensive analysis, the test is only recommendedon suspect transformer on the basis of DGA and othertests. DP is correlated to the tensile strength. Thepaper is considered to have no mechanical strength ata DP of 150 to 200. New transformers have DP above900.

2.3 The following methods of moisturemeasurement are available:

a) Direct measurement of the water content inpaper on a layer by layer basis by means ofKarl Fisher moisture meter (KFMM) using‘four bar probe’.

b) Water content in winding established frommeasurement of tanδ using Gussenbaeur’scharacteristics [4].

c) Water content in winding established frommeasurement of oil PPM using NorrisDiagram.

d) Water content in winding established fromRVM [6].

2.4 Condition Monitoring Methods:

Method Suitable forWINDING:

IR & PI Moisture &Contamination

DDF & Capacitance Moisture &Contamination

Winding Ratio Faulted turnsWinding Resistance Conductor damageImpedance & Losses Winding movementApplied HV Poor dielectricInduced HV & PD Poor dielectricPD ( acoustic &electrical)

Detect and locate PD

Frequency ResponseAnalysis(Transfer Function)

Winding displacement,possible loose winding,and core faults

Voltage Recovery Moisture in paper And ageing of paperOff-Line Test

Vibration Analysis Slack winding &Mechanical faults

DP & Gel Permeation Insulation degradation OIL:

DGA & Ratioanalysis

Detecting incipientfaults

Furan Analysis Overheating andageing of paper

Water content DrynessResistivity, Acidity,IFTand DDF

Ageing of oil

BUSHINGS:Tan Delta (DLA) Moisture &

contamination

Tan δ or dielectric loss angle (DLA) is a measure ofoverall quality of insulation system in terms ofmoisture and contamination.

2.5 Recovery Voltage Method:

If the DGA analysis is performed correctly (propersampling, storage and calibration), most of theincipient faults in the oil may be detected. The paperinsulation is responsible for containing most of themoisture due to ageing and thermal stress. The paperinsulation may fail under high electrical stress or may

release moisture into the oil insulation. To detectageing or moisture content it is necessary to analyselow frequency part of polarisation spectrum ofdissipation factor. A tanδ would have been sufficientbut finding a sinusoidal source voltage of 0.001 Hz isvery difficult [6]. The alternative is the recoveryvoltage measurement.

It was found that IR & PI do not reflect completeinformation on polarisation process. Cases werereported where electrical motors having good PI werefound to have contaminated windings and also motorshaving poor PI had no problems in the windinginsulation [2]. To resolve this, an one thousandseconds charging and discharging test (dc absorption)was developed. Recovery Voltage Method fortransformer seems to be developed from this test.

In RVM, winding is charged for known time and thenshorted to ground for pre-decided time. The recoveryvoltage is then measured and dominant time constantis achieved which is essentially a polarisation timeconstant.

Charging current is given as the sum of displacementcurrent, the polarisation current and the conductioncurrent. Polarising current is dependent on materialproperty and state of ageing. The polarisation ofdielectric can be expressed as sum of various slowpolarisation phenomena like ion migration, slowrelaxation and interfacial polarisation. Care must betaken in the interpretation of results of RVM, inparticular the relative effects of moisture, genuineageing and temperature [3].

2.6 Condition Monitoring Model [1]

Insulation DP of Failure ExpertAgeing Insulation Rate AssessmentProducts Paper DataIn Oil Sample RVM

INSULATION AGE FAILURE PROBABILITY

Remaining Reliable LifeOf Transformer

Data Base OfTest Results

Statistical Analysis

RELIABILITY ANALYSIS

3. REFURBISHMENT DATA

Five 66/11kV, 10-30MVA transformers of 24 --33years old transformers were refurbished. Some of theimportant parameters are given in table 1, 2, 3 & 4.

Table 1

PI % Moisture(from RVM)

% MechanicalStrength (from

DP)1.09 2.52 35—551.79 2.10 35—551.80 1.74 35—552.34 1.56 35—552.40 1.42 25—35

Table 2

% Moistureestimated from

Oil PPM(Norris

Diagram)

% Moistureestimated fromwinding DLA

(Cigre)

% Moistureestimated from

RVM

2 2 1.744 3 1.53

2.5 2 2.523.2 2.8 1.421.5 < 0.5 2.10

Table 3

% Moisture(from RVM) DP

% MechanicalStrength (from

DP)2.09 250—350 35—551.53 250—350 35—551.21 200—250 25—352.38 250—350 35—552.18 250—350 35—55

Norris diagram gives estimated percentage of moisturein paper based upon moisture content in oil (PPM) andoil temperature.

Cigre paper has established a method to estimatemoisture in paper based upon winding tan δ (DLA)and temperature.

Table 4

2-Furfural(PPM) DP

% MechanicalStrength (from

DP)1.10 200—250 25—350.49 250—350 35—550.39 200—250 25—350.54 250—350 35—55

2-Furfurylalcohol, 2-Acetylfuran, 5-Methyl-2-Furfuraland 5-Hydroxymethyl-2-Furfural were not detectedin all four cases.

4. DISCUSSION AND RESULTS

• It is observed from table 1 that RVM tends tofollow PI in this case. However, a conclusioncan not be drawn based on this limitednumber of observations. Essentially RVMdoes have DC absorption phenomenon andmay / may not follow IR / PI.

• It is also observed that DP values (orpercentage retention of mechanical strength)do not correlate with any of the electricalparameters. DP reflects ageing and remainingmechanical strength of paper but it is not adiagnostic test as such.

• From table 3, it is evident that moisturepercentage in the winding does not correlateto DP or mechanical strength retention.

• From table 2, it is evident that percentage ofmoisture estimated from Norris diagram(based upon oil PPM and temperature) andpercentage moisture estimated from Cigrepublication (based upon winding DLA) tendto agree to some extent. However, these donot correlate with moisture percentageestimated from RVM. This is due to the factthat RVM is a total reflection of ageing ofcellulose paper and moisture content in thewinding. There are no established methodsto separate these components.

• It appears from table 4 that Furan analysismay, in some cases, give indication aboutpaper ageing. However, most of the times,Furans are not detected and there is nouniversal correlation available for DP andFurans. Also, both ageing and moisturecontent influence Furans.

5. CONCLUSION

Mechanical integrity of paper is very important toensure reliability of transformers. However at thisstage this is not conclusively reflected in any of theelectrical diagnostic tests available. There is a need forfurther research to validate RVM results and establishmore effective method of estimating moisture contentin the cellulose paper winding.

6.0 REFERENCES:

1. D.ALLAN, “Condition Monitoring & LifeAssessment Of Aged Transmission / Sub-Transmission Plant”, Lecture notes of the shortcourse program on Insulation ConditionMonitoring And Reliability Centred MaintenanceOf Electrical Plant. February 1997, MonashUniversity.

2. CAJETAN PINTO, “An Improved Method ofdetecting Contamination of H.V. Stator Windingin the Field", Proceedings of the 20th ElectricalElectronic Insulation Conference Boston, 1991,pp. 55-59.

3. R.E.JAMES, “Assessment Of ElectricalInsulation”, Proceedings of the 1998 ResidentialSchool in Electrical Power Engineering,Tasmania.

4. I.GUSSENBAUER, “Examination of HumidityDistribution in Transformer Models by Means ofDielectric Measurements”, Cigre paper 15-02,International Conference on Large High VoltageElectric System, 1980.

5. THOMAS LEIBFRIED, “On-line Monitors KeepTransformers in Service”, IEEE ComputerApplication in Power, July 1998, pp 36-42.

6. SYED MOFIZUL ISLAM, “ Power TransformerCondition Monitoring – Pacific PowerInternational and University of NewcastleDevelopments”, Lecture notes of the short courseprogram on Insulation Condition Monitoring AndReliability Centred Maintenance Of ElectricalPlant. February 1997, Monash University.