PARTIAL DISCHARGE LOCATION Selected Topics

1. PARTIAL DISCHARGE LOCATIONSelected Topics M.S. Mashikian IMCORP ICC Meeting - Fall 2000

2. TOPICS • TESTING PHILOSOPHY • Excitation Voltage • Test Voltage Level • METHOD OF REFLECTOMETRY • METHOD OF ARRIVAL TIME • TESTING • Branched Circuits • On-line - Discussion of ON and OFF-Line • Major Concerns - Answers 1

3. TESTING PHILOSOPHY • PURPOSE: • Reduce Service Failures - Improve Reliability • Reduce Cost or Delay Capital Expenditure • TEST GOALS: • Identify defects that may cause near-term service failures • Characterize severity of defects to help prioritize corrective actions • Constraint: Minimize likelihood of test-produced damage • Other 2

4. TESTING PHILOSOPHYEXCITATION VOLTAGE • Use 50/60Hz produced by resonant test set • Ramp voltage slowly until PD is detected or max. test voltage is reached • Dwell approx. 2 seconds (will be reduced to ~ 10 cycles). • Reduce voltage to zero very rapidly. 3

5. RESONANT TRANSFORMER Inductor 1. Variable air-gap inductor 2. Fixed inductor L Cable C Resonance: wL = 1/wC U1 1. U1 is 50/60 Hz 2. U1 is variable frequency L w 4

6. 0° 90° 180° 270° 360° PARTIAL WAVE - 50/60HZ 1/4 or 1/2-cycle 50/60Hz dc PD is detectable only in the quadrants shown 5

7. OSCILLATING WAVE (OSW) • Frequency dictated by cable capacitance • Attenuation depends on cable dc 6

8. APDAC 50/60 Hz + dc - ALTERNATING POLARITY DC-BIASED AC 7

9. TEST VOLTAGE LEVEL • Decide after discussing with client • Minimum: 2U0 recommended • Maximum: 2.5 - 3.0U0 • Rationale Consider maximum transient voltage that may exist on system: geographic location and system protection. 8

10. 100 80 60 40 20 0 1.0 1.3 1.7 2.5 3.0 CUMULATIVE PD PROBABILITY, 2500km, EXTRUDED CABLE Cumulative PD Probability, % PDIV Range, U0 DATA BASED ON 960 PD EVENTS IN CABLES At 1.0 U0, less than 3% of PD events can be identified. 9

11. TEST VOLTAGE LEVEL • PDIV varies with type of excitation voltage: With sinusoidal VLF, PDIV may be as high as 2.5 times that at 50/60Hz. • Where PD is detected with 50/60Hz voltage, no PD may seem to exist at the same voltage level in PD tests with other types of excitation voltage. 10

12. PRINCIPLES OF PD TESTINGPD Site Location • Reflectometry (off-line) • Arrival-Time • Non-branched (2-termination) system • Off-Line • On-Line • Branched system: • Off-Line • On-Line • Frequency-Domain Testing IMCORP Technologies 11

13. 25 t C A B 2 20 ~ x t 1 PD Site 15 Estimator 10 t 1 x AB AC = t 2 5 0 -5 0 2 4 6 8 10 REFLECTOMETRY Amplitude, mV Time, ms 12

14. TEST REPORTJoint and PD Locations PD Joint 13

15. TEST REPORTJoint and PD Locations PD Joint 14

16. PHASE RESOLVED PD DISPLAY PD in Termination 15

17. sensor 1 x x L-x t1 t2 ~ ARRIVAL-TIMENON-BRANCHED CABLE Principle sensor 2 x = 1/2[v(t1-t2)+L] Sensor 1 Actual Sensor 2 Signal Conditioner From SDCT SDCT Multi-Channel Estimator SDCT = Signal detection - conditioning - transmission 16

18. ARRIVAL-TIME BRANCHED CABLES Remote #2 SDCT Test Transformer Remote #3 Sensor Branched Cable Signal conditioner From #1,2,3 SDCT Estimator SDCT Multi-Channel Remote #1 17

19. TESTING 4-TERMINAL CIRCUIT 18

20. PD-LOCATION 19

21. PD TESTING - Major Concerns • Is PD test able to locate water trees? • All trees? • Some trees? See McBride et al. • Does PD testing unduly deteriorate or fail • Good cables? • Defective cables? - Effect of high dc or thumping • How effective is ON-LINE testing? • How effective is PD testing of PILC cables? • Are PD test results obtained by different methods identical? 20

22. McBride et al. (Georgia Power Res., and EPRI) - IEEE Trans. on Power Del. 4/94 70mils 70mils 35x48-mil void 21

23. McBride et al. (Georgia Power Res., and EPRI) - IEEE Trans. on Power Del. 4/94 • Inspection of these wafers revealed a large void (35 mils wide and 48 mils high) in the center of the (70 mils diameter and 70 mils high) water tree. • There was significant evidence of PD within the interior of the water tree, with darkened areas near the tips of the tree. This was clearly the site of the PD activity. • Cable failures near large water trees have occurred in ac breakdown tests… 22

24. McBride et al. (Georgia Power Res., and EPRI) - IEEE Trans. on Power Del. 4/94 • Seven cables failed while active PD sites were monitored (at higher voltage than PDIV). • With the exception of two samples (out of the remaining 7 samples with PD) all other cable failures occurred at the PD inception voltage while (PD site) locations were being performed… Failures were associated with the PD sites. • The duration of sustained PD ranged from 2 to 80 minutes prior to failure. • The theory that PD above operating voltage results in failure at operating voltage is not supported by this data. 23

25. McBride et al. (Georgia Power Res., and EPRI) - IEEE Trans. on Power Del. 4/94 • Very few cables contained active PD sites near operating voltage. No PD above 1-3pC was detected in any of the test cables (61 samples) when energized at operating voltage. (PDIV = 8 - 35kV). • Most PD sites were detected at twice operating voltage or higher. • The (pC) level of the PD at a site could not be correlated with the likelihood of a failure at or near the site. 24

26. BREAKDOWN AT PD SITE BREAKDOWN AT PD SITE THROUGH VENTED WATER TREE THROUGH VENTED WATER TREE 25

27. 0 .6 0 BEFORE UPTI TESTING A FT E R UP TI TE S T I N G 0 .5 0 24 1 Fee d e r Ca b l es 0 .4 0 SAIFI 0 .3 0 0 .2 0 0 .1 0 554 Fe e de r C a b l e s 0 .0 0 1 9 9 7 1 9 9 2 1 9 9 3 1 9 9 4 1 9 9 5 1 9 9 6 1 9 9 8 1 9 9 9 2 0 0 0 Year IMPROVED RELIABILITY Peter Daly, San Francisco, Jan. 20, 2000 26 NOTE: Only 40% of Recommended Repair Was Completed