Page 1
1
Estimating the impact of VLF Frequency on Effectiveness of
VLF Withstand Diagnostics for MV Cable Systems
Nigel Hampton1, Joshua Perkel1, JC Hernandez2, Marina Kuntsevich3, and Vivek Tomer3
1NEETRAC2Universidad de los Andes, Merida
3Dow Chemical
Page 2
2
Outline
• Background• Length Issues • Test Protocols• Water Tree Lengths• Breakdown Strengths• Conclusions
Page 3
3
Introduction
• .
• VLF is used in Simple Withstand, Monitored Withstand and Tan Delta Tests
• A range of frequencies are permitted: 0.01 to 1 Hz
VLF is the waveform most widely used by utilities who employ diagnostic tests
Page 4
Examples of VLF Sources
Page 5
VLF Waveforms
SEBAKMT VLF40 @5kv RMS with 280feet XLPE Load
-8000
-6000
-4000
-2000
0
2000
4000
6000
Volta
ge (k
V)
HVA30 @5kV RMS with 280feet XLPE Load
-8000
-6000
-4000
-2000
0
2000
4000
6000
8000
Volta
ge (k
V)
Sinusoidal
Cosine-Rectangular
Page 6
Preferred Test Protocol
HOLD
RAMP
Time
Voltage
“Real Time decision here if you wish to curtail or
extend Withstand
“Real Time decision here if you wish to continue Withstand
30 Mins
Page 7
VLF Frequency
• A range of frequencies are permitted: 0.01 to 1 Hz
• The frequency changes when the length of the cable system being tested increases – longer cable require lower frequencies
Concern that the withstand result may dependuponfrequency
Malaysia 11 & 33kV System, Moh, CIRED 20030.1Hz 0.05Hz 0.02Hz
Survival 87% 75% 74%Fail On Test (FOT) 10% 19% 20%
Fail In Service (FIS) 3% 6% 6%
Page 8
Lengths Tested with VLF
8
Circuit Length [Conductor ft]
Perc
ent
840007200060000480003600024000120000
30
25
20
15
10
5
0
Median Length = 3500 ft
Page 9
9
Time on Test [Minutes]
Failu
res
onTe
st[%
ofTo
talT
ests
]
100.010.01.00.1
20
10
5
3
2
1
1000 Feet500 FeetNONE
AdjustmentLength
Length Effect on Failures on Test
Time on Test [Minutes]
Failu
res
onTe
st[%
ofTo
talT
ests
]
100.010.01.00.1
20
10
5
3
2
1
1000 Feet500 FeetNONE
AdjustmentLength
Time on Test [Minutes]
Failu
res
onTe
st[%
ofTo
talT
ests
]
100.010.01.00.1
20
10
5
3
2
1
4.1%
2.4%
17.2%
30
1000 Feet500 FeetNONE
AdjustmentLength
Page 10
Length Effects
• Comparison of withstand failure on test rates must include length adjustments
• Lower test frequencies come from the longer lengths
15 kV 0.1Hz L=1
0.05Hz L=2
0.02HzL=5
35 kV 0.02HzL=3
0.1Hz 0.05Hz 0.02HzFail On Test 10% 19% 20%
Length Adj FOT 10% 20 – 30% 30 – 50%
Page 11
• Observed effect is likely not an effect of frequency• Observed effect is likely an effect of length which in turn
impacts frequency
• Is there an effect of frequency?
• To investigate will need 1. consistent defects2. forced frequency
Effect of Frequency
Page 12
Ashcraft Test
• Ashcraft test is a way to reproducibly grow water trees in the laboratory
• Water trees grow from a water needle
• Tree inception & growth are accelerated by
• Field enhancement
at the water needle
• Ionic solution
• High AC frequency
• After 30 days we end up with consistently treed cells that can be VLF tested at selected conditions
Page 13
Test Program
• Grow water trees to consistent lengths in selected materials: EPR, WTRXLPE, XLPE
• Step Test groups (4) to failure using sinusoidal VLF at selected frequencies (0.1 & 0.05Hz)
– Establish VLF strength– Water Tree Length
• If low frequency VLF is less effective then there should be a measurable increase in the VLF breakdown strength
Page 14
Water Tree Lengths very Comparible
40
5
01
51
02
52
0 01 02 03 0
W
tnecreP
)noitalusnI fo %( htgneL eerT reta
0qerF FLV
01.050.
Page 15
VLF Breakdown Strength – water treed
10987654
99
90
8070605040
30
20
10
5
3
2
1
Estimated Mean Breakdown Strength (kV/mm)
Perc
ent
0.050.10
Freq Weibull
Low VLF frequency is not to the right (higher) than common 0.1 Hz VLF
Page 16
Weibull Confidence Limits
Confidence limits overlap for low VLF frequency and common 0.1 Hz VLF
Page 17
Breakdown Strength & Water Trees
100806040200
7.5
7.0
6.5
6.0
5.5
5.0
4.5
4.0
Water Tree Length - Long Trees (% of ins)
Bre
akdo
wn
Stre
ngth
- E
arly
Fai
lure
s (k
V/m
m)
0.05 Hz0.1 Hz
FrequencyVLF
Page 18
Conclusions
• The reported VLF frequency effect on simple withstand is consistent with being due to the increased length of the circuits tested not VLF frequency
• VLF frequency is correlated with increased FOT & FIS but does not cause the effect
• Controlled tests, on very similarly degraded samples, does not show the hypothesised increase in breakdown strength with decreasing frequency
• VLF tests at lower frequencies are likely to be, wrt common 0.1 Hz, either – as effective or – marginally more effective (resulting in lower
breakdown strengths)