A STUDY OF THE EFFECT OF SHATTERED GLASS DISCS …mydocs.epri.com/docs/publicmeetingmaterials/1108/6XNSUMJE9MT/ESKOM... · was contracted by Eskom to conduct electrical breakdown

A STUDY OF THE EFFECT OF SHATTERED GLASS DISCS ONSHATTERED GLASS DISCS ON

INSULATION STRENGTH UNDER HVDC CONDITIONS

SUPPLEMENTAL PROJECT

EPRI HVDC Conference, California 30-31August 2011

Contents

• Background to the project

E i t l S t• Experimental Setup

• Test Program

• Results

• Discussion

• Conclusions

• Recommendations for further work• Recommendations for further work

2011/09/01 2

Background to the project - Introduction

• Eskom currently operates and maintains the ± 533 kV HVDC Cahora Bassa Transmission linebetween Mozambique and South Africa. Live work is limited to maintenance tasks done at half thenominal operating voltage.

f f• Due to high availability requirements and to obtain a better understanding of safety issues related tothe insulation strength of broken glass insulator discs, the Electric Power Research Institute (EPRI)was contracted by Eskom to conduct electrical breakdown tests on glass cap-and-pin insulators asused on the line.

• Glass insulators are prone to vandalism and line inspections have shown that some glass stringshave randomly positioned shattered discs. It is assumed that this condition may affect the totalinsulation strength of the string. The Transmission Grid proposes that live line maintenance work beconducted to repair or replace these discs or strings. At present, live work is not being conducted atfull system voltage and is limited to minor maintenance work. The system voltage is reduced to 75%

f t d lt (400 kV) d t th t i t i i l ti d i t th t i ithof rated voltage (400 kV) due to the uncertainty in insulation and air gap strength on strings withbroken units, and to maintain the recommended safety clearances.

2011/09/01 3

Background to the project – Test Objectives

• The objectives of the series of tests were to gain an understanding of the impact that broken glassdiscs have on the insulation strength of a cap and pin string, with and without the presence ofcorona/space charge. The tests were separated into three main areas:

( f )• Full string insulator breakdown tests (used as the reference case),

• One and Two sequential shattered disc placed at predetermined locations,

• One and Two sequential shattered discs in the presence of additional corona.

• All tests were conducted in accordance with the IEC60060-1 standard, and the results have beencorrected for Standard Temperature and Pressure (STP).

• The tests were conducted at the EPRI High Voltage Laboratory in Lenox Massachusetts, USAg g y ,

2011/09/01 4

Experimental Setup

• REPLICATION OF THE TOWER TOP

In order to conduct experimental tests, it wasp ,decided to closely replicate the physicaldimensions of the predominant suspensiontower.

To properly simulate the tower and conductorconfiguration, an earth plane was manufacturedand mounted approximately 20 feet away fromthe test insulator string The earth plane wasthe test insulator string. The earth plane wasmade of expanded wire mesh within a woodenframe. The dimensions of the frame were 20feet long by 8 feet wide.

In order to simulate the quad-Zambezi conductorbundle, an arrangement of metallic tubing (of asimilar diameter as the actual conductors) wasused

2011/09/01 5

used.

Experimental Setup

• POWER SUPPLY

The power supply was limited to ±1100 kV directcurrent (DC) as a result of the limitation of thewall bushing.

An iron beam measuring 12.2 m was suspendedin the centre of the building using two insulatedwinches. The tower configuration was thenwinches. The tower configuration was thenreplicated in the centre of the beam.

2011/09/01 6

Test Program

Full String Breakdown Tests

• The maximum string length that could practically be flashed over using the ±1100 kVvoltage supply at the Lenox facility was determinedvoltage supply at the Lenox facility was determined.

• The tests were started using a 6 disc string and then increasing the number of insulatordiscs until it was no longer possible to obtain a flashover.

• Apart from determining the maximum number of discs that could be tested, the testsalso assisted in understanding if there was linearity between the number of discs in thestring and the measured flashover voltage.

2011/09/01 7

Test Program (2)

Shattered/Broken Glass disc tests

• The aim of these tests was to determine the effect of shattered discs on the insulationstrength of the insulatorsstrength of the insulators.

• The number of shattered discs, as well as the position of these discs in the insulatorstring were varied.

• The total number of insulators used in this string was determined by the maximumnumber of discs that was flashed over in the previous “Full String Breakdown” tests.The test program involved eight separate tests, namely:

• Test 1: Full string with 1 broken disc at live end,g• Test 2: Full string with 1 broken disc 1/3 up string,• Test 3: Full string with 1 broken disc 2/3 up string,• Test 4: Full string with 1 broken disc at dead end,• Test 5: Full string with 2 broken discs at dead end,• Test 6: Full string with 2 broken discs 2/3 up string,• Test 7: Full string with 2 broken discs 1/3 up string,• Test 8: Full string with 2 broken discs at live end

2011/09/01 8

Test 8: Full string with 2 broken discs at live end.

Test Procedure

• DC voltage was applied to the test string and stepped up in one minute intervals. Thisprocedure was followed until a flashover occurred. After flashover, the room wasvented and the charge on the test object was allowed to drain. The time intervalbetween each test was 30 minutes.between each test was 30 minutes.

• The program listed above was conducted for positive and negative polarity.

• In order to determine the effect that corona and space charge had on the tests, coronap gsources were inserted at the live end only, the dead end only, and simultaneously boththe live and dead ends, for the full string case.

2011/09/01 9

Results

Full String Breakdown Tests

Three tests were conducted on full strings of 6, 10 and 13 discs. The tests indicate that for the casestested, there is a linear relationship between the total flashover voltages of the full strings and the

fnumber of discs in the string.

It is observed that the DC voltage required to flashover a string of 13 discs is +1216 kV and-1110 kV, respectively (corrected for standard temperature and pressure). As expected, the resultsindicate a reduction in flashover voltage for the negative case (by 8.7%).

2011/09/01 10

Results (2)

Shattered discs

A broken disc located at any location in the insulator string will reduce the flashover voltage. Theworst effect is seen for the case where the two shattered discs are located at the live end of the

fstring. This is observed for both polarities.

An unexpected results under negative polarity shows that for the single broken disc case, the lowestbreakdown voltage is achieved when the disc was located at the dead end. This is not consistentwith the positive polarity case. Two broken discs at the live end of the string results in the worstflashover performance for both the Positive and Negative polarity (a 24% reduction in flashovervoltage in both cases).

2011/09/01 11

Conclusions

• The positive 4, 6 and 13 disc flashover results seem to be linear in the region tested,however the assumption cannot be made that the flashover voltage will increase linearlyfor an increase in the number discs beyond what was tested.

• It is concluded that shattered discs results in a definite decrease in insulationstrength.

• As expected, the Negative polarity tests resulted in lower flashover voltages for allp g p y gconfigurations. This represents a situation worse than for positive polarity.

• Generally, the more broken discs at a particular position in the string the lower is thecorresponding flashover voltage. The two exceptions to this were when for the negativepolarit hen the discs ere positioned 2/3 p the string and at the dead endpolarity when the discs were positioned 2/3 up the string and at the dead end.

• The flashover voltage (when broken discs were present in the string) generallyincreased as the broken discs were positioned away from the live end. The exceptionsto this were for the negative polarity when the discs were positioned 2/3 up the stringto this were for the negative polarity when the discs were positioned 2/3 up the stringand at the dead end.

2011/09/01 12

Recommendations

• Although the results have provided detailed information that can be used toimprove Eskom's understanding of the impact of shattered discs on the reductionof insulation strength, further recommendations are made for future studies:

• Corona effects on voltage distribution and flashover voltage will need furtherinvestigation. The tests will need to be carefully planned and undertaken to ensurethat the applied test voltage is more representative of the actual line voltage, i.e. thecorona intensities in the tests should be representative of practical field conditions.An excessively high test voltage will lead to the masking of the corona source effectsAn excessively high test voltage will lead to the masking of the corona source effects.

• Since the tests were performed on clean insulators the following further studies arerecommended:

• The effect of pollution on insulator flashover voltage,• The effect of pollution on insulator voltage distribution,• The combination of the effect of polluted insulator strings and shattered glass

discs.• Full string flashover voltage linearity needs to be studiedFull string flashover voltage linearity needs to be studied.• Electric field modelling of the environment around the insulator will be considered, in

the next phases.

2011/09/01 13

Thank youProject Team MembersEskom: Project Leader : Nish Parus - [email protected] specialist : Nishal Mahatho - [email protected] Engineer : Thavi Govender - [email protected] Line Specialist : Eric Marshall - [email protected] Line Manager : Eric Greyling - [email protected] Consultant (HVDC) : Piet Goosen - [email protected]

EPRI: Gary SibilantGeorge Gela D RDave RuegerChris MooreMike Clary