on-site partial discharge measurement of transformers introduction

"ON-SITE PARTIAL DISCHARGE MEASUREMENT OF TRANSFORMERS by Victor Sokolov, Vladimir Mayakov ZTZ-Service Co (Ukraine)

Georgy Kuchinsky,

Alexander Golubev

Technical University, Sankt-Peterburg (Russia)

Cutler-Hammer (USA)

INTRODUCTION During recent years the technical policy of power utilities is changing under the pressure of economical considerations. There is a common tendency of moving from the time-based to the condition-based maintenance, to reduce maintenance costs, and, at a time, to prevent of sudden failure, particularly catastrophic one that typically follows to insulation fault. A vital problem is also how to provide continuity of operation of the questionable transformer especially if that shows symptoms of PD activity. Then, one may emphasize a clear tendency to perform maintenance procedures infield condition, especially without de-energizing the equipment. Question arises how to verify quality of such works? Incipient failure of insulation attributes always to PD activity. PD measurement [1, 2, 3, 4] is recognized as the main and the most reliable method of evaluation of the insulation condition at the factories. Why it may be not duplicated in field, particularly to verify repair quality and to ensure that incipient fault suggested e.g. by DGA test will not progress to affect the insulation system? There have been several technical obstacles, especially high external noise and needs in special expensive facilities. Therefore, only system of acoustical PD detection has been practically used, in some instances very effectively [5, 6]. The recent progress in the PD measuring technique has opened really new opportunities in an effective rejection of external interference, and on-site diagnosis of the condition of transformer insulation, quite similarly to well established tests at the transformer factories.. Successful implementation of PD electrical test technique and its apparent advantages has been reported in several papers [7, 8, 9, 10, 18]. This paper discusses problems associated with PD activity and PD identification in Power Transformers, and effectiveness of diagnostic methods available. The case histories with identification of problems in transformers using PD Analyzer are presented.

1

HISTORY ZTZ-Service Co has utilized PD measurement technique since early 70th. At first, it was an attempt to duplicate in the field of factory PD measurements. Large Electrostatic shields placed on the bushings and separated source of energizing have been used to minimize external noise. However, feasible PD signal could have been achieved on the level of 5,000-10,000 pC only [12]. Special balance (bridge) circuits has allowed reducing PD signal to 600-900 pC (tests of shunt reactors [14]), and even to 50-100 pC using separate diesel-generator as a source of energizing [13]. Such tests due to high costs have been performed occasionally, as a part of life extension program. Preferably, acoustic monitoring technique with piezoelectric transducers have been utilized, basically on the units exhibiting PD or arcing activity through DGA test. Since 1998, ZTZ – Service has implemented Electric Partial Discharge Analyzer .The practical experience has confirmed that this test technique provides high enough sensitivity to PD in field conditions and may be used as an effective diagnostic instrument [11]. The following transformers have been considered as candidates for PD monitoring: • Exhibiting PD or arcing activity through DGA test • After Repair and Refurbishment in-field • Ranking the units requiring in repair • Assessment of maintenance (processing) program • Assessment the condition of critical transformers (nuclear units, equipment of 750 kV transmission system, etc.)

Partial Discharge problems in Power Transformers. A View from Experience The Table 1 shows classification of PD generation sources in operating transformers, which has been defined on the basis of many years of experience. There have been three potential sources of PD generation in Power Transformers: “Core and Coils Assembly”, Bushings, and LTC. On the other hand the sources of PD can be associated with operative voltage, with voltage induced by main magnetic flux, and with voltage induced by stray flux. Statistics has shown that the problems associated with sparking and arcing within a transformer could be roughly distributed as the following: PD associated with main magnetic flux 31 % PD associated with stray flux 41 % PD associated with operative voltage Conductor under floating potential, bad contacts 14 % Creeping discharge 14 % The most dangerous cases with PD activity in major insulation spaces make up less then 15 % of the total problems associated with arcing inside a transformer. In such case the unit must have been taken from operation immediately. On the other hand, failure analysis has shown [11] that about 15-20 % of annual sudden failures and most of catastrophic events are caused by impairment of the conditions of main and minor insulation due to a particle contamination or the ingress of moisture reducing the

2

impulse withstands strength. These events definitely attributed to PD activity frequently remain undetected.

MECHANISMS OF INSULATION DEGRADATIONINSULATION DETERIORATION

Defect-free transformer insulation is characterized with apparent PD magnitude 10-50 pC or lower. Increasing the PD level up to 100-300 pC is associated typically with presence of particles and (especially just after filling the transformer with oil) with trapped small air bubbles [16]. Ionization level 25-500 pC [17,19] in an oil barrier space does not affect dielectric withstand strength and may be considered as characteristic of normal deterioration. Defected condition. Defected but reversible change of insulation condition is associated with PD>1000pC, and may be caused by different degradation mechanisms: • Conductive mode particles, which bridge oil gap and result in discharges with magnitude from 100 to 10,000pC[16]. • Increasing the moisture content in a paper up to 3-4 % and relevant increasing concentration of moisture in oil, which causes reduction of PD inception voltage by 20 % and occurrence of PD’s with the level up to 2000-4000 pC[20 ]. • Poor impregnation may cause discharges of about 1,000-2,000 pC [16]. • Large (3-5 mm) air/gas bubble in oil may result in discharges ranging in magnitude from 1,000 to 10,000pC [16]. In general, PD level over 2500pC (in paper) and over 10,000 pC (in oil) may be considered as destructive ionization [1,19] in a long-term action.

3

Table 1 Sources of PD generation Components Core and Coils Assembly oil-barrier-paper structure and oil

Source of PD Operating voltage PD attributed to reversible change of insulation condition PD attributed to irreversible degradation of insulating material:

Electrostatic Shields Leeds

PD associated with voltage induced by main magnetic flux

PD associated with voltage induced by stray flux

Oil/surface contamination with particles, bubbles, static electrification, bad impregnation, high moisture, Partial breakdown in oil; surface discharge; creeping discharge Sparking and arcing between bad connection “Conductor under floating potential” discharges Tracking in wooden blocks. closed loops between adjacent members linked y the main flux (insulated bolts of core, pressing bolts, pressing metal rings, etc); sparking due to floating potential closed loops between adjacent members linked by stray flux; floating potential (e.g. ungrounded magnetic shunts)

Operating voltage

Localized defect within the core: bad impregnation, high moisture, short-circuits between layers, sparking across the core surface Breakdown in oil; surface discharge across the porcelain

Operating voltage PD associated with operating voltage at the fix tap position: PD associated with switching process

Partial breakdown in the selector and in the divertor switch compartment

Bushings

LTC

Typical faults

Poor or worn out contacts

4

Mechanism of Incipient irreversible failure in oil-barrier insulation: • • • • • • • •

Initiated by breakdown of oil gap which is registrated as apparent charge >10,000 pC which rise rapidly to 100,000-1000,000pC[20,21]. Progresses in surface discharge in oil across the barrier with PD magnitude over 100,000 pC. Appearance of “white marks” on the surface due to forcing oil and water out of the pressboard pores followed with carbonized “black mark” on the barrier. PD 10,000-100,000 may cause irreversible damage during tens of hours. Minimum energy required to cause incipient carbonizing the cellulose (heating over 300C) is estimated as 0.1 J, which corresponds to several charge pulses of 100,000-1000,000 pC [1], Stable charge in oil is associated with PD power P>0.4 W. average rate of gas generation under effect of stable PD in oil- 50µl/J[22,23]. Further steps progress either in breakdown of the insulation space or in occurrence of creeping discharge.

Creeping discharge That is likely the most dangerous failure mode that typically results in catastrophic failures at normal operating conditions. The phenomenon occurs in the composite oil-barrier insulation and progresses in several steps: • Partial breakdown of oil gap. • Surface discharge in oil across a barrier (an appearance of black carbonized mark on the barrier). • Microscopic sparking within the pressboard. The presence of some excessive moisture stimulates this process. • Splitting oil molecules under effect of sparking. Formation of hydrocarbons followed with the formation of carbonized traces in the pressboard. This process is resulted in lowering PD magnitude of apparent charge to 1,000-5,000 pC. Creeping process can continue from minutes to months or even years, until the treeing conductive path causes shunting of an essential part of transformer insulation resulting in a powerful arc. Cellulose destruction (creeping discharge) while progressing within the pressboard is associated with PD intensity q>1000 p C; PD power P=0.1-1 W, Average rate of gas generation of 40-50 µl/Joule [23]. Failure of turn-to turn insulation • • •

Occurrence of sporadic PD pulses of magnitude over 400 –1000 pC during several hours [24]. Rapid rise PD activity up to q>100,000pC (duration up to tens of minutes). Paper puncture, turn-to-turn short circuit (duration – several second).

5

Diagnostic characteristics of a faulty transformer insulation Faulty conditions can be characterized in terms of PD activity, fault gas generation, and changes in the conductance, capacitance and dielectric loss factor of a defective area. Typical scenario of initially defect-free insulation failure is the following: Contamination (particles, water, bubbles) Occurrence of moderate PD Occurrence of destructive PD Gas generation Progressing PD, accompanied with gas generation Tracking/treeing, accompanied with gassing and changing dielectric characteristics, critical pre-failure PD Breakdown Relevant diagnostic characteristics are: • PD parameters: apparent charge magnitude, Pulse repetition rate, Discharge Power, PD Signature. • Faulty gas content. • Gas generation rate as a rate of degradation of insulating materials, which is a function of discharge power. • Change of Power Factor, Conductivity and Capacitance of the defective insulation space [25]. Diagnostic technique shall advise how to distinguish between really dangerous problems (e.g. destructive PD in the oil-barrier structure), and problems, which does not affect the functionality of a transformer to a danger extent, and an equipment could be kept in service at least for some while. Apparently, consideration about possible critical transformer condition must be supported by design review, especially to assess “the sensitive points of the equipment. Methods of PD registration Three of most frequently used types of PD sensing are electric, acoustic and electromagnetic. Con and Pro of the methods are presented in the Table 2. Ideally, the combination of all three methods can be a powerful diagnostic tool: • Rough detection of the problem externally using electromagnetic sensor, • Identification of insulation condition with electric method, • Location of the PD source by an acoustic device. However, the only electric method can detect and identify defective condition that associated with PD activity ranged below 1,000 pC. Our experience has confirmed that acoustic detection of PD is a very effective complementary tool to localize a source of gassing caused by very strong arcing in oil. Diagnostic method in this case can be: DGA-is a trigger and acoustic sensor-is a locating tool. However, the cases, when identification of arcing problems through “DGA + acoustic” technique is effective, are associated with the dissipation of high energy (typically, over 500-1000 kJ) and show clear only strong degradation process. Acoustic PD monitoring has not been effective in identifying of progressing creeping discharge. Minimum PD signal effectively detecting by acoustic sensor in full-scale transformer model is limited to the level of approximately 10,000 pC and above.

6

Table 2 Methods of PD registration TYPE OF SENSORS

ADVANTAGES

ELECTRIC Direct connection to the test tap, or through high frequency CT on the grounded wire, (“Rogovski coils”) Additional sensors in bus duct, electrostatic shields, neural etc.

High sensitivity Can be calibrated in terms of apparent charge Approximate location of PD source All capabilities to trend data Use of PD pattern Recognition technology Sensors configuration can match for better noise rejection Easy to use Possible assessing external PD problems including PD in the bushings Serving as a noise (corona) channel Easy to install Capability detecting acoustic emission magnitude and trend, Pulse repetition rate and trend Localizing a source of PD using signal time of arrival to different locations

ELECTROMAGNETIC (ANTENNA)

ACOUSTIC Piezo-accelerometer placed on transformer tank

DISADVANTAGES De-energizing for sensors installation

High disturbances Only discharges of extremely high level can be detected Difficult to distinguish an equipment having problems from surrounding equipment Low sensitivity Minimal detecting apparent charge >10,000pC Responded to rain, sleet, electrical disturbances in the station The level of signal depends on coupling between the sensor and surface Effect of design (variables inside the tank) on the propagation of sound wave.

Experience with Partial Discharge identification using Universal PD Analyzer PD Technology The portable Universal PD Analyzer UPDA “Twins” has been used for different infield applications to detect PD activity in Power and Instrument transformers. The UPDA monitors the wave forms for several power frequency cycles from several (up to 8) sensors simultaneously, and then identifies the PD pulses in the acquired information using up to 5 independent noise deletion systems [26]. Radio frequency pulses are measured at the output of the capacitance taps of the bushings (typically six sensors), neutral and tank/active part grounding wires. Some of measuring inputs is used for connection of acoustic sensors and electromagnetic sensor (antenna).

7

Reduction of noise is provided by the following measures: • Selection of frequency band. As a rule, the frequency band from 1 MHz to 20 MHz is selected. UPDA also has the option to choose full frequency band or to use an external filter. • Filtering (sometimes, when radio transmission interference is very high). • Electronic processing of signals: phase and frequency selection, time delay selection, signal comparison, etc. • Time window method (provided with the gate, which can be open and closed at pre-selected moments). • Comparison of signals from different electromagnetic sensors (to reduce corona disturbances, in CT’s specifically). • Balanced circuit- in the shunt reactors and CT’s. High frequency band used in measurements creates some problems associated with attenuation of PD signals, while they travelling in a transformer to a sensor. Therefore, a calibration is important to evaluate sensitivity and to determine possible impact to a data by attenuation. High frequency band may also exert an effect on screening PD impulses in the range of 100-1,000 kHz, which typically accompany of powerful PD 100,000-1000,000 pC. Tests procedure can be adjusted for the particular transformer design. Calibration is an important process, which is typically provided immediately after the monitoring system commissioning. Calibration allows for: • Recalculate pulse magnitude into an apparent charge. • Evaluate sensitivity of the each measuring channel. • Compose calibrating cross-matrix, which serve as an important diagnostic tool, especially to distinguish between PD sources in the core, in the bushings or in winding insulation, and to determine an approximate PD location. Test Protocol includes cross-matrix, detected noise level, residual (white) noise level, maximum pulse magnitude, Pulse Repetition rate, and PD Power. Threedimensional phase-resolved analysis is also available for PD pattern recognition. Diagnostic method consists of several steps: • Evaluating PD characteristics. • Considering voltage/load/temperature effects on PD activity. • Considering accompanied factors (DGA, oil tests, and dielectric characteristics). • Design review. Comparative tests at the factories The effectiveness of noise deletion by UPDA Analyzer was evaluated by comparative tests in the factory. The tests using factory technique and UPDA were performed on the transformer 500/220 , 100 MVA (Table 3) and on the Current Transformer 330 kV (Table 4). Factory test setup uses all possible noise reduction means including screened room and disel-generator as a power source. The tests have shown a good agreement between factory Lab’s and UPDA technique. It was found that UPDA Analyzer achieves satisfactory signal to noise ratio, even without special procedures, which are critical for factory setup.

8

Table 3 Comparative PD test on the 100 MVA, 500/220 transformer Place of sensor Applied voltage

Factory’s test technique

UPDA Analyzer

Special shied on 500kVbushing

Without shield

Special shield on 500kV bushing

Without shield

24pC

2800 pC

30.7pC

68.7pC

HV ,neutral, Grounding 476 kV,1.64 Ur

Table 4 PD test on the 330 kV current transformer Place of sensor Applied voltage

Factory’s test technique Balanced circuit

UPDA Analyzer Direct test

26 pC

11pC

Measured tap, Grounding tape ( «0» – electrod) 275 V (1.3 Ur.)

In Field experience. Case of History. Case 1

Verification the quality of in-field repair.

Auto-transformer 200MVA, 330/110 kV has failed after 13 years in operation due to explosion of 330 kV bushing. The failure was accompanied with the tank rupture, damage to HV winding insulation and severe contamination of the core and coils assembly with carbon, pieces of porcelain etc. ZTZ-Service has performed remedy repair on the substation in-field conditions. Insulation condition after repair was evaluated as satisfactory. (Water content in the pressboard 0.7 %, PF =0.2% at 45 C, low content of particles in oil, etc.). However, some doubt remained especially referring to possible localized contamination in HV winding. PD measurement has been performed with UPDA Twins system at rated voltage and at 1.05 Ur. The unit was energized from 110kV side. Test results (Table 5) have confirmed a good condition of the insulation integrity.

9

Table 5 Customer: Object: Sensor

PD Test Report , Date:12-18-1998 200.0/330/110, 122580, -2 Deleted Max PD Magnitude Repetition Noise, [nC] Rate,[ppc] [V] [pC] Bush_A_S 0,989 0,0266 53.2 1,11 Bush_B_S 3,42 0,0376 75.2 1,88 Bush_C_S 0,547 0,0298 59.7 2,07 Bush_Am_S 4,63 0,0266 47.9 1,01 Bush_Bm_S 4,68 0,178 320 12,3 Bush_Cm_S 0,66 0,075 135 11,1 Neutr_S 0,169 0,0133 40 10,4

PD Power,[mW] 0,531 0,904 0,838 0,294 16,7 9,06 1,11

Case 2 Assessing of the condition of 220 kV transformer after long service The PD tests have been incorporated in the program of life assessment of the 240MVA, 220/110 kV, core-form auto-transformer with the goal to make a decision about urgent necessity of insulation reconditioning after 28 years of service. Water Heat Run Test (WHRT) has revealed symptoms of significant insulation aging. The heating of the transformer up to 65C and holding for 24 hours has shown doubling of water-in-oil content, reduction in dielectric breakdown voltage by 20 % and increasing its variation coefficient to 15 % (symptom of presence of large particles). Particles in-oil count has shown the level of contamination denominated as “High”(NAS class 7). Water content in cellulose estimated through PF test and WHRT was about 2 %. Table 6 Dissolve Gas Analysis results for the 240 MVA transformer H2

CH4

4.7

.

C2H2 .

C2H4 52

C2H6 .

CO

CO2

162

1388

Σ C3 67

2

1-C4H8 2039

N2

0.60 2.21

10

Table 7 Calibrating cross-matrix Calibrating impulse A_220 -Tank B_220 – Tank C_220 – Tank Am_110 – Tank Bm_110 – Tank Cm_110 – Tank Neutral – Tank Tank - Ground _220 –Tap _220 – Tap _220 – Tap Am_110 – Tap Bm_110 – Tap m_110 - Tap

Sensors inputs A_220 1.000 0.231 0.081 0.054

B_220 0.054 1.000 0.135 0.076

C_220 0.076 0.212 1.000 0.043

Am_110 0.120 0.077 0.054 1.000

Bm-110 0.033 0.269 0.095 0.207

Cm_110 0.065 0.096 0.135 0.065

0.022

0.044

0.061

0.050

1.000

0.033

0.011

0.011

0.073

0.164

0.145

0.145

0.255

1.000

0.036

0.036

0.058 0.143 1.000 0.290 0.091 0.046 0.022 0.105

0.050 0.143 0.069 1.000 0.109 0.062 0.033 0.063

0.058 0.214 0.069 0.290 1.000 0.031 0.033 0.053

0.033 0.071 0.139 0.161 0.073 1.000 0.050 0.074

0.075 0.107 0.056 0.516 0.127 0.115 1.000 0.126

0.067 0.143 0.069 0.194 0.164 0.069 0.044 1.000

1.000 0.393 0.028 0.065 0.036 0.015 0.011 0.021

0.017 1.000 0.028 0.065 0.036 0.015 0.011 0.021

Table 8 PD Test Report Object:TDCTGA-240000/220, Location:SB Baltijskaya

Neutral 0.022 0.038 0.027 0.022

Ground 0.022 0.038 0.027 0.022

79463, 2 , data17.09.1999

Sensor

PD Power [mW]

A_220_S B_220_S C_220_S Am_110_S Bm_110_S Cm_110_S Neutral_S Ground_S

84.5 53.1 88.3 20.5 9.39 1.38 20.1 0.14

Residual noise [nC] 0.074 0.128 0.078 0.068 0.036 0.115 0.057 0.031

Max. Pulse Magnitude [nC] 0.298 0.355 0.389 0.127 0.160 0.066 0.303 0.0426

Repetition Rate [ppc] 150.0 128.0 132.0 95.1 31.7 7.35 32.4 0.466

PD tests data are shown in the Tables 6,7,8, including calibrating cross-matrix, the level of residual (white) noise and parameters of PD. The results have shown a moderate deterioration of the insulation integrity. Maximum PD magnitude in the insulating spaces having minimal safety margin (insulation of the series winding 220kV) was below 400pC. Correspondingly, it was advised postponing a repair of the unit and prolong operation of transformer without oil and insulation processing.

11

Case 3. Assessment of seriousness of gas generation in 300 MVA ,500 kV transformer One of two sisters –type shell-form autotransformers shown clear generation of faulty gases, likely associated with combination of PD activity , localized oil heating ,and cellulose decomposition Continuing of operation of the both units was very critical. The problem arose how serious is the symptom of abnormality? Will it progress to affect the insulation system? Table9 DGA in oil data for the 300 MVA transformer Object

H2

Data

CH4

C2H6

C2H4

C2H2

CO

CO2

[ppm] T1RS

27.09.97

1

1

4

1

1

4

49

T1RS

17.10.97

53

2

1

1

2

46

235

T1RS

12.11.97

154

6

2

3

2

105

684

T1RS

14.05.98

447

32

11

23

1

580

1920

T1RS

30.06.98

563

151

50

148

1

605

1757

T1RS

23.07.98

557

155

45

156