cdfi icc education spring 2009 minutes - used slides & … · 2009. 7. 16. · 6 21 presented...
TRANSCRIPT
1
1Presented at the Spring 2009 ICC Education Session Copyright GTRC 2009
ICC Education Session
Cable Diagnostic Focused Initiative
Nigel HamptonRick Hartlein
Joshua Perkel
Spring 2009 Meeting
2Presented at the Spring 2009 ICC Education Session Copyright GTRC 2009
GTRI/DOE Disclaimer• The information contained herein is to our knowledge accurate and reliable at
the date of publication. • Neither GTRC nor The Georgia Institute of Technology nor NEETRAC will be
responsible for any injury to or death of persons or damage to or destruction of property or for any other loss, damage or injury of any kind whatsoever resulting from the use of the project results and/or data. GTRC, GIT and NEETRAC disclaim any and all warranties both express and implied with respect to analysis or research or results contained in this report.
• It is the user's responsibility to conduct the necessary assessments in order to satisfy themselves as to the suitability of the products or recommendations for the user's particular purpose.
• No statement herein shall be construed as an endorsement of any product or process or provider
• Any opinions, findings and conclusions or recommendations expressed in this material are those of the author(s) and do not necessarily reflect the views of the Department of Energy
• This material is based upon work supported by the Department of Energy under Award No DE-FC02-04CH1237
3Presented at the Spring 2009 ICC Education Session Copyright GTRC 2009
Outline
• CDFI Background/Overview • Cable System Failure Process• SAGE Concept• Case Study: Roswell• Diagnostic Accuracies• Diagnostic Testing Technologies• Accuracies Really Matter• The Things We Know Now That We Did Not Know Before• Selecting a Diagnostic Testing Technology• Summary
4Presented at the Spring 2009 ICC Education Session Copyright GTRC 2009
CDFI Background
Rick Hartlein
2
5Presented at the Spring 2009 ICC Education Session Copyright GTRC 2009
• Underground cable system infrastructure is aging (and failing). Much of the system is older than its design life.
• Not enough money / manufacturing capacity to simply replace cable systems because they are old.
• Need diagnostic tools that can help us decide which cables/accessories to replace & which can be left in service.
• Always remember that we are talking about the cable SYSTEM, not just cable.
Cabl
e Fa
ilure
s pe
r Y
ear
20052000199519901985198019751970
1000
800
600
400
200
0
Why do we need diagnostics?
CDFI Background/Overview 6Presented at the Spring 2009 ICC Education Session Copyright GTRC 2009
Composition of US MV system
Inst
alle
d Ca
paci
ty (
%)
UNKNOWNTRXLPEEPRXLPEHMWPEPILC
100
90
80
70
60
50
40
30
20
10
0
25
50
75
CDFI Background/Overview
7Presented at the Spring 2009 ICC Education Session Copyright GTRC 2009
Failure Split
Unknown1.1%Terminations
5.6%
Splices37.1% Cable
56.2%
CDFI Background/Overview 8Presented at the Spring 2009 ICC Education Session Copyright GTRC 2009
• In the CDFI, NEETRAC worked with 17 utilities, 5 manufacturers and 5 diagnostic providers to achieve the objective of clarifying the concerns and defining the benefits of diagnostic testing.
• Phase 1 has almost exclusively focused on aged medium voltage systems.
• This is the largest coherent study of cable system diagnostics anywhere.
Overview
CDFI Background/Overview
3
9Presented at the Spring 2009 ICC Education Session Copyright GTRC 2009
NEETRAC Members
Non NEETRACMembers Supporters
Dept of Energy
Diagnostic Providers
CDFI
CDFI Background/Overview 10Presented at the Spring 2009 ICC Education Session Copyright GTRC 2009
Participants
SouthwireSouthern CompanySouthern California EdisonTyco / Raychem Public Service Electric & GasPrysmianOncor (TXU)PEPCOPacific Gas & Electric (added Jan 06)PacifiCorp (added mid 2005)NRECAIMCORPHydro QuebecHV Technologies
CenterPoint Energy
GRESCO
HV Diagnostics
Cablewise / Utilx
Florida Power & Light
Con Edison
HDW Electronics
Georgia Tech
First EnergyExelon (Commonwealth Edison & PECO)
Duke Power CompanyCooper Power Systems
AmerenAmerican Electric Power
CDFI Background/Overview
11Presented at the Spring 2009 ICC Education Session Copyright GTRC 2009
CDFI - Primary Activities1) Technology Review2) Analysis of Existing (Historical) Data3) Collection and Analysis of Field (New) Data4) Verification of VLF Test Levels5) Defect Characterization6) Develop Knowledge Based System7) Quantify Economic Benefits8) Reports, Update Meetings and Tech Transfer
Seminars
Analyses are data / results driven
CDFI Background/Overview 12Presented at the Spring 2009 ICC Education Session Copyright GTRC 2009
CDFI Activities
CDFI
Analysis Lab Studies
Field Studies Dissemination
CDFI Background/Overview
4
13Presented at the Spring 2009 ICC Education Session Copyright GTRC 2009
CDFI Activities
CDFI
Analysis Lab Studies
Field Studies Dissemination
Value / Benefit
Accuracies
Utility Data
IEEE Std Work
VLF Withstand
Tan δ
PD
Georgia Power
Duke
Handbook
Publications
Meetings
Industry
CDFIKnowledge Based Systems
CDFI Background/Overview 14Presented at the Spring 2009 ICC Education Session Copyright GTRC 2009
CDFI Activities
Lab Studies
VLF Withstand Tan δ PD
Test TimeTest VoltageForensics
Time StabilityVoltage Stability
Non-Uniform DegradationNeutral Corrosion
CalibrationPhase Pattern
Feature ExtractionClassification
CDFI Background/Overview
15Presented at the Spring 2009 ICC Education Session Copyright GTRC 2009
CDFI ActivitiesField
Studies
Georgia Power XLPE
Jkt & UnJkt21 Conductor Miles
DukeXLPE & Paper
Jkt & UnJkt29 Conductor Miles
Offline PD (0.1Hz)Offline PD (60Hz)
Tan δMonitored Withstand
Offline PD (0.1Hz)Tan δ
Monitored Withstand
Charlotte * 2CincinnatiClemson
Morresville
EvansMaconRoswell
CDFI Background/Overview 16Presented at the Spring 2009 ICC Education Session Copyright GTRC 2009
CDFI Activities
Analysis89,000 Conductor Miles
Value / Benefit Accuracies UtilityData IEEE Std Work Knowledge
Based Systems
Economic ModelSAGE
DC WithstandOffline PDOnline PD
Tan δVLF Withstand
400 Omnibus400.2 VLF
SurveyExpert System
Application
CDFI Background/Overview
5
17Presented at the Spring 2009 ICC Education Session Copyright GTRC 2009
CDFI Activities
UtilityData
Con Ed Com Ed PPL Alabama Power Keyspan
DC WithstandOnline PD
VLF Withstand
Offline PD (60Hz)Online PDTan Delta
VLF Withstand
Offline PD (0.1Hz)Tan Delta Online PD Offline PD (0.1Hz)
Tan Delta
CDFI Background/Overview 18Presented at the Spring 2009 ICC Education Session Copyright GTRC 2009
CDFI Activities
UtilityData
FPL
Offline PD (60Hz)VLF Withstand
PEPCO
Offline PD (60Hz)Offline PD (0.1Hz)
Online PDVLF Withstand
PG&E ONCOR Ameren
Offline PD (60Hz)Online PD
Tan δ
Offline PD (60Hz)Online PD Offline PD (60Hz)
CDFI Background/Overview
19Presented at the Spring 2009 ICC Education Session Copyright GTRC 2009
Dataset Sizes
89,000ALLService Performance
Diagnostic
Data Type
-0.3IRC
9,8101.5VLF Withstand
5501.5Tan δ
262-PD Online
4902PD Offline
149-Monitored Withstand
78,105-DC Withstand
Field[Conductor miles]
Laboratory[Conductor miles]Technique
CDFI Background/Overview 20Presented at the Spring 2009 ICC Education Session Copyright GTRC 2009
Time [Days]
Log
Cum
ulat
ive
Failu
res
3000200015001000900800
1500
1000
500
100
Time [Days]
Log
Cum
ulat
ive
Failu
res
3000200015001000900800
1500
1000
500
100
Benefits from Diagnostic ProgramsDecreasing failures associated with diagnostics and actions
CDFI Background/Overview
Program Initiated
6
21Presented at the Spring 2009 ICC Education Session Copyright GTRC 2009
At the Start• For many utilities, the usefulness of diagnostic testing was
unclear.
• The focus was on the technique, not the approach.
• The economic benefits were not well defined.
• There was almost no independently collated and analyzed data.
• There were no independent tools for evaluating diagnostic effectiveness.
CDFI Background/Overview 22Presented at the Spring 2009 ICC Education Session Copyright GTRC 2009
Where we are today (1)1. Diagnostics work – they tell you many useful things, but not
everything.2. Diagnostics do not work in all situations.3. Diagnostics have great difficulty definitively determining the
longevity of individual devices. 4. Utilities HAVE to act on ALL replacement/repair
recommendations to get improved reliability.5. The performance of a diagnostic program depends on
• Where you use the diagnostic• When you use the diagnostic• What diagnostic you use• What you do afterwards
CDFI Background/Overview
23Presented at the Spring 2009 ICC Education Session Copyright GTRC 2009
6. Quantitative analysis is complex BUT is needed to clearly see benefits.
7. Diagnostic data require skilled interpretation to establish how to act.
8. No one diagnostic is likely to provide the detailed data required for accurate diagnoses.
9. Large quantities of field data are needed to establish the accuracy/limitations of different diagnostic technologies.
10. Important to have correct expectations – diagnostics are useful but not perfect!
CDFI Background/Overview
Where we are today (2)
24Presented at the Spring 2009 ICC Education Session Copyright GTRC 2009
• In the CDFI, NEETRAC worked with 17 utilities, 5 manufacturers and 5 diagnostic providers to achieve the objective of clarifying the concerns and defining the benefits of diagnostic testing.
• We have come a long way wrt the project objective. – Analysis driven by data / results– Developed a good understanding that diagnostic testing can
be useful, but the technologies are not perfect.– Developed ways to define diagnostic technology accuracy and
found ways to handle inaccuracies. – Developed diagnostic technology selection and economic
analysis tools.– Understand that there is yet more to learn.
Overview
CDFI Background/Overview
7
25Presented at the Spring 2009 ICC Education Session Copyright GTRC 2009
How things fail and what fails have a big impact on the selection of diagnostics
Cable System Failure ProcessRick Hartlein
26Presented at the Spring 2009 ICC Education Session Copyright GTRC 2009
Failures by Equipment
Unknown - all (%)Terminations - all (%)Splice - all (%)Cable - all (%)
100
80
60
40
20
0
Dis
burs
emen
t of
Fai
lure
s (%
)
Cable System Failure Process
27Presented at the Spring 2009 ICC Education Session Copyright GTRC 2009
Major Cable Components
Jacket (Recommended)
Metallic Shield/Neutral
Insulation Shield
Insulation
Conductor or Strand ShieldConductor
Cable System Failure Process 28Presented at the Spring 2009 ICC Education Session Copyright GTRC 2009
1. Cavity at shield(s)2. Cavities due to shrinkage3. Insulation shield defect4. Contaminant (poor adhesion)5. Protrusions at shield(s)6,7 Splinter/Fiber8. Contaminants in insulation or shields
Defect Types in Extruded Cables
Cable System Failure Process
8
29Presented at the Spring 2009 ICC Education Session Copyright GTRC 2009
Conversion of Water to Electrical Trees
• Acts as a stress enhancement or protrusion (non-conducting)
• Water tree increases local electric field
• Water tree also creates local mechanical stresses
• If electrical and mechanical stresses high enough ⇒electrical tree initiates
• Electrical tree completes the failure path – rapid growth
Electrical tree growing from water tree
Cable System Failure Process 30Presented at the Spring 2009 ICC Education Session Copyright GTRC 2009
Defect Types in Extruded Cable Accessories
Cable System Failure Process
31Presented at the Spring 2009 ICC Education Session Copyright GTRC 2009
Diagnostics used in Challenging Areas
32Presented at the Spring 2009 ICC Education Session Copyright GTRC 2009
Summary• Cable system aging is a complex phenomenon.
• Multiple factors cause systems to age.
• Increases in dielectric loss and partial discharge are key phenomenon.
• The aging process is nonlinear.
• Diagnostics must take these factors into consideration.
Cable System Failure Process
9
33Presented at the Spring 2009 ICC Education Session Copyright GTRC 2009
SAGE Approach to
Diagnostic Programs
Nigel Hampton
34Presented at the Spring 2009 ICC Education Session Copyright GTRC 2009
Diagnostic Program Phases - SAGESelectionData compilation and analysis needed to identify circuits that
are at-risk for failure (at-risk population).
ActionDetermine what actions can be taken on circuits based on the
results of diagnostic testing.
GenerationConduct diagnostic testing of the at-risk population.
EvaluationMonitor at-risk population after testing to observe/improve
performance of diagnostic program.SAGE Concept
35Presented at the Spring 2009 ICC Education Session Copyright GTRC 2009
SAGE at WorkFailures [#]
Time
Selection
Action
Generation
Evaluation
Decreasing Failures
Increasing Failures
SAGE Concept 36Presented at the Spring 2009 ICC Education Session Copyright GTRC 2009
When to deploy diagnostics
Time (Years)
Cabl
e Sy
stem
Per
form
ance
403020100
Operational Stress
Condition AssessmentCommissioning
10
37Presented at the Spring 2009 ICC Education Session Copyright GTRC 2009
Global ContextComparison with many tests
DatabasesStandards
Context – is important
Local ContextComparisons within one area
DataGeneration from
Diagnostic Measurement
SAGE Concept 38Presented at the Spring 2009 ICC Education Session Copyright GTRC 2009
Case StudyRoswell, GA
November 2008 & January 2009
Nigel Hampton
TDRTan Delta
Monitored WithstandOffline PD
39Presented at the Spring 2009 ICC Education Session Copyright GTRC 2009
39
Roswell Map
Case Study: Roswell 40Presented at the Spring 2009 ICC Education Session Copyright GTRC 2009
SELECTION
Case Study: Roswell
11
41Presented at the Spring 2009 ICC Education Session Copyright GTRC 2009
Roswell Background Info.• 1980 vintage XLPE feeder cable, 1000 kcmil, 260 mils wall,
jacketed.
• Failures have occurred over the years – no data on source
• Recently experienced very high failure rates of splices on this section: 80 failures / 100 miles / yr.
• Overall there have been 10 -15 failures of these splices in last two years on a variety of GPC feeders.
• Splice replacement may be acceptable if there is a technical basis.
Case Study: Roswell 42Presented at the Spring 2009 ICC Education Session Copyright GTRC 2009
Knowledge Based Selection System
Selecting a Diagnostic Technology
43Presented at the Spring 2009 ICC Education Session Copyright GTRC 2009
Summary for Diagnostic Selection
Replace AccessoriesReplace SegmentReplace Small Portion
TDR
& H
istorical R
ecords ON
LY
PD O
ffline
PD O
nline
Tan Delta
Monitored
Withstand
HV D
C Leakage
VLF 60 Mins
VLF 30 Mins
VLF 15 Mins
DC
Withstand
Diagnostic Technique
ActionScenario
Have a shortlist of three techniques
Case Study: Roswell 44Presented at the Spring 2009 ICC Education Session Copyright GTRC 2009
Economic Details – prior to testing• Complete System Replacement $1,000,000 approx• Complete Splice Replacement $60,000• Test time (determined by switching) 3 - 4 Days• Selection Costs $5,000• Splice Replacement 7 Days• Retest after remediation 1 Day
Monitored Withstand, Offline PD and VLF (30 mins) offer economic benefit over doing nothing.
Case Study: Roswell
12
45Presented at the Spring 2009 ICC Education Session Copyright GTRC 2009
Scenario Assessment before TestingOffline PD• If 51,000ft is tested• 0.5% fails on test, no customer
interrupted • 1 site / 1,000ft (median)• 40% discharges in cable• Estimate
– 0 fails on test– 51 discharge sites
• 20 cable, • 31 accessories
– 15 splices– <2 failure in 12 months from
test
Monitored Withstand• If 51,000ft is tested• <4% fails on test, no customer
interrupted• 70% of loss tests indicate no
further action• Estimate
– <2 fails on test– 3 assessed for further
consideration by loss – 0.5 failure in 12 months
from test
46Presented at the Spring 2009 ICC Education Session Copyright GTRC 2009
ACTION
Case Study: Roswell
47Presented at the Spring 2009 ICC Education Session Copyright GTRC 2009
Initial Corrective Action Options
• Replace splices only – no detailed records assume 12 splices.
• Complete system replacement.
Case Study: Roswell 48Presented at the Spring 2009 ICC Education Session Copyright GTRC 2009
GENERATION
Case Study: Roswell
13
49Presented at the Spring 2009 ICC Education Session Copyright GTRC 2009
Overhead and Cabinet Terminations
Case Study: Roswell 50Presented at the Spring 2009 ICC Education Session Copyright GTRC 2009
Monitored Withstand
Case Study: Roswell
51Presented at the Spring 2009 ICC Education Session Copyright GTRC 2009
If this had been a Simple Withstand
Length Tested (miles)1086420
18 Segments Tested
No Failures On Test
Case Study: Roswell 52Presented at the Spring 2009 ICC Education Session Copyright GTRC 2009
Monitored Withstand - Stability
Sequence of Lengths Tested (miles)1086420
18 Segments Tested
Case Study: Roswell
Sequence of Lengths Tested (miles)1086420
Pass - Un Stable Loss
Pass - Stable Loss
18 Segments Tested
60 min test
30 min test
14
53Presented at the Spring 2009 ICC Education Session Copyright GTRC 2009
Test Results - Local Perspective
Length Along Feeder (ft)
Tip
Up in
Tan
Del
ta {
1.5U
o -
0.5U
o} (
1e-3
) 1000
100
10
1
150001000050000
150001000050000
1000
100
10
1
1 2
3
STABLEUNSTABLE
StabilityMeasurement
7
6
53
2
1
76
53
2
1
76
5
3
21
Panel variable: Phase
Segment ID'sNumbers indicate
Case Study: Roswell 54Presented at the Spring 2009 ICC Education Session Copyright GTRC 2009
Test Results – Global Perspective
Tip Up 1.5Uo - 0.5 Uo (1e-3)
Tan
Del
ta @
Uo
(1e-
3)
1000100101
100.0
10.0
1.0
0.1
1 150
6
150STABLEUNSTABLE
Stability
ErrorSplice
withstandmonitoredinstability inRange of
Termination Damage
Case Study: Roswell
55Presented at the Spring 2009 ICC Education Session Copyright GTRC 2009
Targeted Offline PD
Case Study: Roswell 56Presented at the Spring 2009 ICC Education Session Copyright GTRC 2009
Targeted Offline PD Test – Segment 6
Distance from Cubicle 2 (ft)
Phase
PD
TDR
C - 3
B - 2
A - 1
C - 3
B - 2
A - 1
5000450040003500300025002000150010005000
Positions)(Approx
Distance from Cubicle 2 (ft)
Phase
PD
TDR
C - 3
B - 2
A - 1
C - 3
B - 2
A - 1
5000450040003500300025002000150010005000
Positions)(Approx
anomalous TDR reflectionsOpen symbols represent the
Distance from Cubicle 2 (ft)
Phase
PD
TDR
C - 3
B - 2
A - 1
C - 3
B - 2
A - 1
5000450040003500300025002000150010005000
Positions)(Approx
anomalous TDR reflectionsOpen symbols represent the
Case Study: Roswell
15
57Presented at the Spring 2009 ICC Education Session Copyright GTRC 2009
PD Inception – local perspective
Position from Cubicle 2 (ft)
VLF
Tes
t V
olta
ge (
kV)
232119171513
10
5
048003600240012000
48003600240012000
232119171513
10
5
0
A - 1 B - 2
C - 3
3133
2126
36372126
4088
1681 785
Panel variable: Phase
PD in 1 of 9 splices PD in 1 of 7 splices
PD in 5 of 9 splices
Position of PD (ft)
PD Inception (kV)
Prob
abili
ty o
f Sp
lice
Ince
ptio
n (%
)
201510987
90
80
706050
40
30
20
10
5
3
2
1
Case Study: Roswell 58Presented at the Spring 2009 ICC Education Session Copyright GTRC 2009
EVALUATION
Case Study: Roswell
59Presented at the Spring 2009 ICC Education Session Copyright GTRC 2009
Evaluation after TestingOffline PD• 15,000ft actually tested• Estimate
– 15 discharge sites • 6 cable, • 9 accessories
– 6 splices– <1 failure in 12 months from
test• Actual
– 7 discharge sites • 0 cable,• 7 accessories
– 25 splices– 0 failure 4 months from
test
Monitored Withstand• 51,000ft actually tested• Estimate
– 2 fails on test– 3 assessed for further
consideration by loss – 0.5 failure in 12 months
from test
• Actual– 0 fails on test– 6 assessed for further
consideration by stability, tip up & loss
– 1 failure (cable) 5 months from test
60Presented at the Spring 2009 ICC Education Session Copyright GTRC 2009
After Testing…• Actions have been performed by GPC.
– Suspect splice investigated, actually broken neutral.– Damaged termination replaced.– Test excavations & Ground Penetrating Radar tests
conducted, concluded that it was not practical to replace splices as planned
• System Re enforcements Planned.
• All tested circuits have been left in service and are being monitored by GPC.
Case Study: Roswell
16
61Presented at the Spring 2009 ICC Education Session Copyright GTRC 2009
Diagnostic Accuracies
Nigel Hampton
62Presented at the Spring 2009 ICC Education Session Copyright GTRC 2009
Performance of Diagnostics
• Performance evaluation primarily focuses on diagnostic accuracy.
• Diagnostic accuracies quantify the diagnostic’s ability to correctly assess a circuit’s condition.
• Accuracy must be assessed based on “pilot” type field test programs in which no actions are performed.
• Circuits must be tracked for a sufficient period of time.
Diagnostic Accuracies
63Presented at the Spring 2009 ICC Education Session Copyright GTRC 2009
Objective of Diagnostic TestsThe target population contains both “Good” and “Bad” components
– “Good” – Will not fail within diagnostic time horizon– “Bad” – Will fail within diagnostic time horizon
“Bad” Components “Good” ComponentsTarget Population
Diagnostic Accuracies 64Presented at the Spring 2009 ICC Education Session Copyright GTRC 2009
Diagnostic OperationApplying the diagnostic will separate the population into:• No Action Required group• Action Required group
But the diagnostic is imperfect...
Diagnostic Accuracies
17
65Presented at the Spring 2009 ICC Education Session Copyright GTRC 2009
Perspective• Diagnostics make measurements in the field and find
Anomalies.• Detecting the presence of an Anomaly is, in our view, not
sufficient.• The goal, in our view, is to detect an Anomaly which leads to
reduced reliability (failure in service) or compromised performance (severed neutrals – stray voltage).
In accuracy estimates we have used failures in service and interpreted the diagnostics as “Bad Means Failure.”
Diagnostic Accuracies 66Presented at the Spring 2009 ICC Education Session Copyright GTRC 2009
“Bad Means Failure” Accuracies
1716151413121110987654321
100
80
60
40
20
0
1716151413121110987654321
No Action Accuracy
Dataset
Dia
gnos
tic
Acc
urac
y
Action Accuracy
1716151413121110987654321
100
80
60
40
20
0
1716151413121110987654321
No Action Accuracy
Dataset
Dia
gnos
tic
Acc
urac
y
Action Accuracy
Diagnostic Accuracies
1716151413121110987654321
100
80
60
40
20
0
1716151413121110987654321
No Action Accuracy
Dataset
Dia
gnos
tic
Acc
urac
y
Action Accuracy
67Presented at the Spring 2009 ICC Education Session Copyright GTRC 2009
Overall AccuracyNo Action AccuracyAction Accuracy
100
80
60
40
20
0
Dia
gnos
tic
Acc
urac
y [%
]
Overall AccuracyNo Action AccuracyAction Accuracy
100
80
60
40
20
0
Dia
gnos
tic
Acc
urac
y [%
]
All Accuracies
Overall AccuracyNo Action AccuracyAction Accuracy
100
80
60
40
20
0
Dia
gnos
tic
Acc
urac
y [%
]
Diagnostic Accuracies
Overall AccuracyNo Action AccuracyAction Accuracy
100
80
60
40
20
0
Dia
gnos
tic
Acc
urac
y [%
]
Overall AccuracyNo Action AccuracyAction Accuracy
100
80
60
40
20
0
Dia
gnos
tic
Acc
urac
y [%
]
68Presented at the Spring 2009 ICC Education Session Copyright GTRC 2009
Diagnostic Testing Technologies
Nigel Hampton
18
69Presented at the Spring 2009 ICC Education Session Copyright GTRC 2009
Introduction• A wide range of diagnostic techniques are commercially
available.
• Tests are performed either offline (circuit de-energized)) or online (energized) and by service providers or utility crews.
• Different voltage sources may be used to perform the same measurement.– DC– 60 Hz. AC– Very Low Frequency (VLF) AC– Damped AC (DAC)
Diagnostic Testing Technologies 70Presented at the Spring 2009 ICC Education Session Copyright GTRC 2009
Utility Use of Diagnostics
Diagnostic Testing Technologies
71Presented at the Spring 2009 ICC Education Session Copyright GTRC 2009
Diagnostic Survey
• A survey of CDFI participants in 2006 was conducted to determine how diagnostics were employed.
• Survey was updated at the end of 2008.
• Survey results focused CDFI work on technologies currently used in the USA.
Diagnostic Testing Technologies 72Presented at the Spring 2009 ICC Education Session Copyright GTRC 2009
No TestingTesting - one techniqueTesting - > one technique
27.8%
30.6%
41.7%
Survey of Use of Diagnostics
19
73Presented at the Spring 2009 ICC Education Session Copyright GTRC 2009
Survey of Use of Diagnostics
Diagnostic Testing Technologies
More than one technique usedNo testingO ne technique used
No TestingOccasional useRegularly usedSome testing
4.0%
96.0%
75.0%
25.0%
No Testing
Testing
74Presented at the Spring 2009 ICC Education Session Copyright GTRC 2009
Lengths Tested
Cable Length - log (ft)
Perc
ent
10.0
7.5
5.0
2.5
0.0100000100001000100
100000100001000100
10.0
7.5
5.0
2.5
0.0
PD Tan D
VLF Withstand
Panel variable: Technique
Median 814 ftMedian 485 ft
Median 3500 ft
Based on diagnostic data supplied to CDFI
Diagnostic Testing Technologies
75Presented at the Spring 2009 ICC Education Session Copyright GTRC 2009
Global ContextComparison with many tests
DatabasesStandards
Context
Local ContextComparisons within one area
DataGeneration from
Diagnostic Measurement
SAGE Concept 76Presented at the Spring 2009 ICC Education Session Copyright GTRC 2009
Diagnostic Context
OK Not Proven either way
NOT OK
• Extreme conditions are easy to decide what to do about.
• What to do about the ones in the middle?
• How to define the boundaries?
Diagnostic Testing Technologies
20
77Presented at the Spring 2009 ICC Education Session Copyright GTRC 2009
Simple Dielectric WithstandTest Description• Application of voltage above normal operating voltage for a
prescribed duration.• Attempts to drive weakest location(s) within cable segment to
failure while segment is not in service.
Field Application• Offline test that may use:
– DC– 60 Hz. AC– VLF AC– Damped AC
• Testing may be performed by a service provider or utility crew.
Diagnostic Testing Technologies 78Presented at the Spring 2009 ICC Education Session Copyright GTRC 2009
Withstand Test Process
HOLDINITIAL
Time
Voltage
t = 0 tTest
Voltages and Times for VLF covered in IEEE 400.2
The goal is to have circuit
out of service, test it such that
“imminent”service failures
are made to occur on the
test and not in service
Diagnostic Testing Technologies
79Presented at the Spring 2009 ICC Education Session Copyright GTRC 2009
VLF Test Voltages
Cable Rating (kV)
Test
Vol
tage
(kV
)
302520151050
60
50
40
30
20
10
0302520151050
Cosine-rectangular Sinusoidal
Peak Voltage (kV) Acceptance
Peak Voltage (kV) InstallationPeak Voltage (kV) MaintenanceRMS Voltage (kV) Acceptance
RMS Voltage (kV) InstallationRMS Voltage (kV) Maintenance
Variable Use
Diagnostic Testing Technologies 80Presented at the Spring 2009 ICC Education Session Copyright GTRC 2009
DataGeneration from
Diagnostic Measurement
Diagnostic Testing Technologies
21
81Presented at the Spring 2009 ICC Education Session Copyright GTRC 2009
Test Sequences
Diagnostic Testing Technologies
Cumulative Length Tested in One Year (Miles)
Wit
hsta
nd T
est
Out
com
es
140120100806040200
26
20
22
22
26
23
16
26
2730
Time of failure in mins for failures > 15 mins
Simple VLF Withstand to IEEE400.2 Levels
82Presented at the Spring 2009 ICC Education Session Copyright GTRC 2009
Local ContextComparisons within one area
Diagnostic Testing Technologies
83Presented at the Spring 2009 ICC Education Session Copyright GTRC 2009
Area
Failu
res
on T
est
[% o
f Te
sted
]
4321Overall
35
30
25
20
15
10
Separation with Simple VLF Outcomes
Area 1 is clearly different from the others.
84Presented at the Spring 2009 ICC Education Session Copyright GTRC 2009
Global ContextComparison with many tests
DatabasesStandards
Diagnostic Testing Technologies
22
85Presented at the Spring 2009 ICC Education Session Copyright GTRC 2009
Withstand Testing Experience
Time on Test [Minutes]
Surv
ivor
s [%
of
Tota
l Len
gths
Test
ed]
706050403020100
100
80
60
40
20
0
IEEE Recommendation
IEEE 400.2 Range
Diagnostic Testing Technologies
9700 Conductor Miles>2000 Conductor Miles
0.3 Conductor Miles
86Presented at the Spring 2009 ICC Education Session Copyright GTRC 2009
Test Performance for Different Utilities
Time on Test [Minutes]
Failu
res
on T
est
[% o
f 10
00 f
t Se
gmen
ts]
100.0010.001.000.100.01
10
5
3
2
1
0.01
4.4%5.0%
0.5%
5.7%
60
A1A2DI
Utility
1000ft Length Adj.
Diagnostic Testing Technologies
87Presented at the Spring 2009 ICC Education Session Copyright GTRC 2009
Service Experience
10000100010010
30
20
10
5
3
2
1
0.1
Time to Failure [Days since test]
Serv
ice
Failu
res
[% o
f To
tal T
este
d]
5%
472
Days
637
Days
10%
1247
Day
s
2247
Day
s
15 Min @ 2.5U030 Min @ 1.8U0
Diagnostic Testing Technologies
2650 Conductor Miles
224763730 Min @ 1.8 U0
472
Time to Failure5%
[Days]124715 Min @ 2.5 U0
Time to Failure10%
[Days]Test Conditions
88Presented at the Spring 2009 ICC Education Session Copyright GTRC 2009
Dielectric Loss (Tan δ)Test Description• Measures total cable system loss (cable, elbows, splices & terminations).• May be performed at one or more frequencies (dielectric spectroscopy).• May be performed at multiple voltage levels.• Monitoring may be conducted for long durations.
Field Application• Offline test that may use:
– 60 Hz. AC– VLF AC– Damped AC
• Testing may be performed by a service provider or utility crew.
Diagnostic Testing Technologies
23
89Presented at the Spring 2009 ICC Education Session Copyright GTRC 2009
Dielectric Loss (Tan δ)Dielectric losses - Tan δ:
V
I
RI CI
1tan( ) R
C
IDFI RC
δω
= = =
VRI
ICI
δ
θ
• The cable insulation system is represented by an equivalent circuit
• In its simplest form it consists of two parameters; a resistor and a capacitor [IEEE Std. 400]
• When voltage is applied to the cable, the total current will be the contributions of the capacitor current and the resistor current
Diagnostic Testing Technologies 90Presented at the Spring 2009 ICC Education Session Copyright GTRC 2009
Cable System Equivalent
Cable system (cable, splices, and terminations) is reduced to simple circuit.
Cable System Failure Process
91Presented at the Spring 2009 ICC Education Session Copyright GTRC 2009
DataGeneration from
Diagnostic Measurement
Diagnostic Testing Technologies 92Presented at the Spring 2009 ICC Education Session Copyright GTRC 2009
Tan δ Test Data
Time [min]
Tan-
delt
a [1
e-3]
543210
100
90
80
70
60
50
40
30
20
10
0.51.01.51.7
[p.u.]Voltage
Time [min]
Tan-
delt
a [1
e-3]
543210
100
90
80
70
60
50
40
30
20
10
0.51.01.51.7
[p.u.]Voltage
Time [min]
Tan-
delt
a [1
e-3]
543210
100
90
80
70
60
50
40
30
20
10
0.51.01.51.7
[p.u.]Voltage
Mean
could be used)Standard Deviation - IQRScatter (represented by
Tip Up
Time [min]
Tan-
delt
a [1
e-3]
543210
100
90
80
70
60
50
40
30
20
10
0.51.01.51.7
[p.u.]Voltage
MeanTip Up
Diagnostic Testing Technologies
24
93Presented at the Spring 2009 ICC Education Session Copyright GTRC 2009
Local ContextComparisons within one area
Diagnostic Testing Technologies 94Presented at the Spring 2009 ICC Education Session Copyright GTRC 2009
Tan δ Data for EPR Cable Systems
Voltage (kV)
Tan
Del
ta (
1e-3
)
1211109876543
25
20
15
10
5 ConcernLowest
ConcernHighest
Diagnostic Testing Technologies
95Presented at the Spring 2009 ICC Education Session Copyright GTRC 2009
Segments within a Feeder
Length Along Feeder (ft)
Tip
Up in
Tan
Del
ta {
1.5U
o -
0.5U
o} (
1e-3
)
1600014000120001000080006000400020000
1000
100
10
1
STABLEUNSTABLE
Stable
7
6
5
3
2
1
Phase = 1
96Presented at the Spring 2009 ICC Education Session Copyright GTRC 2009
Lengths within a Locality
Length (ft)
Tan
Del
ta (
1e-3
)
50004000300020001500
10
1
25
97Presented at the Spring 2009 ICC Education Session Copyright GTRC 2009
Global ContextComparison with many tests
DatabasesStandards
Diagnostic Testing Technologies 98Presented at the Spring 2009 ICC Education Session Copyright GTRC 2009
Testing at Reduced Voltages
Tan-delta @ 2.0 Uo [1E-3]
Tan-
delt
a @
1.5
Uo
[1E-
3]
100.010.01.00.1
100.0
10.0
1.0
0.1
1.2 2.2 4
0.7
1.3
2.3
Regression95% CI95% PI
PI: Prediction IntervalCI: Confidence Interval
“Cool Wall”
99Presented at the Spring 2009 ICC Education Session Copyright GTRC 2009
Tan δ Interpretation
Tip Up
Tan
Del
ta
-1010-1-
-
150
6
0
No Action
Further Study
Action Required
Diagnostic Testing Technologies
Based on 258 Conductor Miles
100Presented at the Spring 2009 ICC Education Session Copyright GTRC 2009
Time Domain Reflectometry (TDR)Test Description• Measures changes in the cable impedance as a function of
circuit length by observing the pattern of wave reflections.• Used to identify locations of accessories, faults, etc.
Field Application• Offline test that uses a low voltage, high frequency pulse
generator.• Testing may be performed by a service provider or utility crew.
Diagnostic Testing Technologies
26
101Presented at the Spring 2009 ICC Education Session Copyright GTRC 2009
TDR Principles
Near End
TDREquipment
Far End
JointL
Joint
102Presented at the Spring 2009 ICC Education Session Copyright GTRC 2009
Online Partial DischargeTest Description• Measurement and interpretation of discharge and signals on
cable segments and/or accessories.• Signals captured over minutes / hours.• Monitoring may be conducted for long durations.
Field Application• Online test that does not require external voltage supply.• Testing typically only be performed by a service provider.• Assessment criteria are unique to each embodiment of the
technology
Diagnostic Testing Technologies
103Presented at the Spring 2009 ICC Education Session Copyright GTRC 2009
DataGeneration from
Diagnostic Measurement
Diagnostic Testing Technologies 104Presented at the Spring 2009 ICC Education Session Copyright GTRC 2009
Discharge Occurrence
No PD PD
Diagnostic Testing Technologies
27
105Presented at the Spring 2009 ICC Education Session Copyright GTRC 2009
Local ContextComparisons within one area
Diagnostic Testing Technologies 106Presented at the Spring 2009 ICC Education Session Copyright GTRC 2009
Distribution of PD along Lengths
Diagnostic Testing Technologies
• 5000 ft. portion of sample feeder
• Mixture of different PD levels for different sections and accessories.
Cable Section Accessory
No PDPD
107Presented at the Spring 2009 ICC Education Session Copyright GTRC 2009
Global ContextComparison with many tests
DatabasesStandards
Diagnostic Testing Technologies 108Presented at the Spring 2009 ICC Education Session Copyright GTRC 2009
Diagnostic Results (Overall)
Accessory Cable
52.5%
42.9%
314.8%
266.6%
113.1%
52.3%
41.8%
314.4%
268.0%
113.4%
Diagnostic Testing Technologies
226 Conductor Miles
28
109Presented at the Spring 2009 ICC Education Session Copyright GTRC 2009
Offline Partial DischargeTest Description• Measurement and interpretation of partial discharge signals
above normal operating voltages.• Signal reflections (combined with TDR information) allows
location to be identified within cable segment.
Field Application• Offline test that may use:
– 60 Hz. AC service provider – VLF AC utility crew– Damped AC utility crew
Diagnostic Testing Technologies 110Presented at the Spring 2009 ICC Education Session Copyright GTRC 2009
PD
P1 P2 P3
Time P1-P3
Time P2-P3
Am
plitu
de [m
V]
-50
10152025
5
0 2 4 6 8Time [μs]
Near End
DetectionEquipment
Far EndP1P2
P3
PD Activity
Diagnostic Testing Technologies
111Presented at the Spring 2009 ICC Education Session Copyright GTRC 2009
DataGeneration from
Diagnostic Measurement
Diagnostic Testing Technologies 112Presented at the Spring 2009 ICC Education Session Copyright GTRC 2009
PD Pulse
140 mV
180 pC
Diagnostic Testing Technologies
29
113Presented at the Spring 2009 ICC Education Session Copyright GTRC 2009
PD Phase Resolved Pattern
3601800
0
-2.50E-1
-1.25E-1
1.25E-1
-323
-162
162
3232.50E-1
0
Phese [Deg]
Am
plitude [pC]
114Presented at the Spring 2009 ICC Education Session Copyright GTRC 2009
PD Magnitude
PD Measurement Voltage (Uo)
PD L
evel
(pC
)
2.252.001.751.50
40
35
30
25
20
15
10
5
the fieldMeasurement fromIndividual
PD Measurement Voltage (Uo)
PD L
evel
(pC
)
2.252.001.751.50
40
35
30
25
20
15
10
5 Max allowed for current production
the fieldMeasurement fromIndividual
PD Measurement Voltage (Uo)
PD L
evel
(pC
)
2.252.001.751.50
40
35
30
25
20
15
10
5 Max allowed for current production
Max Limit for 1970's production
the fieldMeasurement fromIndividual
115Presented at the Spring 2009 ICC Education Session Copyright GTRC 2009
Local ContextComparisons within one area
Diagnostic Testing Technologies 116Presented at the Spring 2009 ICC Education Session Copyright GTRC 2009
PD Charge Magnitude Distributions
Apparent Charge Magnitude [pC]
Perc
ent
6005004003002001000
20
15
10
5
0
Diagnostic Testing Technologies
Apparent Charge Magnitude [pC]
Perc
ent
6005004003002001000
20
15
10
5
0
XLPE
30
117Presented at the Spring 2009 ICC Education Session Copyright GTRC 2009
PD Inception Voltage
Apparent Inception Voltage [U0]
Perc
ent
2.42.11.81.51.20.9
18
16
14
12
10
8
6
4
2
0
Apparent Inception Voltage [U0]
Perc
ent
2.42.11.81.51.20.9
18
16
14
12
10
8
6
4
2
0
XLPE
118Presented at the Spring 2009 ICC Education Session Copyright GTRC 2009
Global ContextComparison with many tests
DatabasesStandards
Diagnostic Testing Technologies
119Presented at the Spring 2009 ICC Education Session Copyright GTRC 2009
119
Location of PD
Termination26.3%
Splice34.3%
Cable39.4%
60.6% of PD sites detected in accessories
Diagnostic Testing Technologies
222 Conductor Miles
120Presented at the Spring 2009 ICC Education Session Copyright GTRC 2009
Offline PD Test Sequence• Testing sequence for 16,000 ft.
No PD
PD
31
121Presented at the Spring 2009 ICC Education Session Copyright GTRC 2009
PD Location
Location [% of Circuit Length]
Perc
ent
9075604530150
18
16
14
12
10
8
6
4
2
0
Terminations
Cable & Splices
122Presented at the Spring 2009 ICC Education Session Copyright GTRC 2009
PD Sites per Length
PD S
ites
per
100
0 fe
et
5
4
3
2
1
Approx. 1 PD Site/1000 ftMedian = 0.96 PD Sites/1000 ft
Diagnostic Testing Technologies
123Presented at the Spring 2009 ICC Education Session Copyright GTRC 2009
Isothermal Relaxation CurrentTest Description• Measures the time constant of trapped charges within the
insulation material as they are discharged.• Discharge current is observed for 15-30 minutes.
Field Application• Offline test that uses DC to charge the cable segment up to
1kV.• Testing is performed by a service provider.
Diagnostic Testing Technologies 124Presented at the Spring 2009 ICC Education Session Copyright GTRC 2009
Recovery VoltageTest Description• Similar to IRC only voltage is monitored instead of current
Field Application• Offline test that requires initial charging by DC source up to
2kV.• Testing is performed by a service provider.
Diagnostic Testing Technologies
32
125Presented at the Spring 2009 ICC Education Session Copyright GTRC 2009
Combined Diagnostics
Multiple degradation mechanisms mean that two diagnostics are often better than one
Diagnostic Testing Technologies 126Presented at the Spring 2009 ICC Education Session Copyright GTRC 2009
No TestingTesting - one techniqueTesting - > one technique
27.8%
30.6%
41.7%
Survey of Use of Diagnostics
127Presented at the Spring 2009 ICC Education Session Copyright GTRC 2009
Multiple Diagnostics
Tan Delta / PDVLF / Tan Delta
Category
75.0%
25.0%
Diagnostic Testing Technologies 128Presented at the Spring 2009 ICC Education Session Copyright GTRC 2009
DataGeneration from
Diagnostic Measurement
Diagnostic Testing Technologies
33
129Presented at the Spring 2009 ICC Education Session Copyright GTRC 2009
Monitored Withstand - Data
Time [min]
Tan-
delt
a [1
e-3]
20151050
1.8
1.7
1.6
1.5
1.4
1.3
1.2
1.1
1.0
1.01.51.7
[p.u.]Voltage
vs. Voltagegives Tan-deltaOptional Ramp Up
through the 15 minute withstandStability of Tan-delta monitored
Diagnostic Testing Technologies 130Presented at the Spring 2009 ICC Education Session Copyright GTRC 2009
Monitored Withstand Data - Elbow
Time [min]
Tan-
delt
a [1
e-3]
76543210
180
160
140
120
100
80
60
40
20
0
0.51.01.51.7
[p.u.]Voltage
Tan-delta vs Voltage
Tan-delta Stability
Before Failure
Failure
After Failure
Segment HL_23_22Elbow Failure Hampton Leas
“Cool Wall”
131Presented at the Spring 2009 ICC Education Session Copyright GTRC 2009
Global ContextComparison with many tests
DatabasesStandards
Diagnostic Testing Technologies 132Presented at the Spring 2009 ICC Education Session Copyright GTRC 2009
Monitored Withstand
Cumulative Length Tested in One Year (Miles)
Wit
hsta
nd T
est
Out
com
es
9080706050403020100
U
NST
ABL
E
H
igh
Loss
Hig
h TU
Poor
Sta
bilit
y
27305
Monitored VLF Withstand to IEEE400.2 Levels
Time of failure in mins
IEEE400.2 LevelsSimple VLF Withstand to
Teste
34
133Presented at the Spring 2009 ICC Education Session Copyright GTRC 2009
Accuracies Revisited
Why do they matter?
Josh Perkel
134Presented at the Spring 2009 ICC Education Session Copyright GTRC 2009
Consequence
Diagnostic Program Costs Cost [$]
Selection
Diagnostic
CorrectiveActions Total Diagnostic
Program Cost
Accuracies Really Matter
135Presented at the Spring 2009 ICC Education Session Copyright GTRC 2009
Recall the Example...
No Action Required Action Required
Avoided Corrective Actions
Avoided service failures
Accuracies Really Matter 136Presented at the Spring 2009 ICC Education Session Copyright GTRC 2009
Incorrect Diagnosis
Future service failures Unneeded Corrective Actions
No Action Required Action Required
Accuracies Really Matter
35
137Presented at the Spring 2009 ICC Education Session Copyright GTRC 2009
Benefit and LossCost [$]
Selection
CorrectiveActions
Diagnostic
ConsequenceAlternate
Program 1
AlternateProgram 2
BENEFIT
LOSS
Accuracies Really Matter 138Presented at the Spring 2009 ICC Education Session Copyright GTRC 2009
Considerations
• Diagnostic program economic calculations are based on ability to predict future failures.
• Total diagnostic program cost is more sensitive to certain elements than others.– Failure Rate– Diagnostic Accuracy– Failure Consequence
Accuracies Really Matter
139Presented at the Spring 2009 ICC Education Session Copyright GTRC 2009
Diagnostic Accuracy Complications
• Time is a critical factor in the assessment of accuracy.– Failures do not happen immediately after testing.
• Two approaches to computing diagnostic accuracy.– “Bad Means Failure” Approach– “Probabilistic” Approach
Accuracies Really Matter 140Presented at the Spring 2009 ICC Education Session Copyright GTRC 2009
Accuracy Over Time – “Bad Means Failure”
Time[Years]2 4 6 108
Accuracy[%]
100
0
No Action Required Accuracy
Action Required Accuracy?
• System Changes• Additional Aging• Increased Load
Accuracies Really Matter
36
141Presented at the Spring 2009 ICC Education Session Copyright GTRC 2009
Probabilistic Approach – Tan δ
100101FOT
5
2
1
0.01
Elasped Time at Oct 08 (Month)
Failu
res
in S
ervi
ce [
% o
f Te
sted
Seg
men
ts]
ActNo Action
Action_1
100101FOT
5
2
1
0.01
Elasped Time at Oct 08 (Month)
Failu
res
in S
ervi
ce [
% o
f Te
sted
Seg
men
ts]
ActNo Action
Action_1
100101FOT
5
2
1
0.01
Elasped Time at Oct 08 (Month)
Failu
res
in S
ervi
ce [
% o
f Te
sted
Seg
men
ts]
ActNo Action
Action_1
Accuracies Really Matter
100101FOT
5
2
1
0.01
Elasped Time at Oct 08 (Month)
Failu
res
in S
ervi
ce [
% o
f Te
sted
Seg
men
ts]
12
7.4%
0.9%
ActNo Act
Action_1
142Presented at the Spring 2009 ICC Education Session Copyright GTRC 2009
Probabilistic Approach - PD
Days Between Test & Failure
Perc
ent
1000100101
99
9080706050403020
10
532
1
0.01
No PDPD
PD Class
Accuracies Really Matter
143Presented at the Spring 2009 ICC Education Session Copyright GTRC 2009
The Things We Know Now That We Did Not Know Before
144Presented at the Spring 2009 ICC Education Session Copyright GTRC 2009
By Diagnostic TechniqueVLF DC Tan Delta
PD On PD Off TDR
IRC DAC
No UseOccasionalStandardTesting
Category
Diagnostic Testing Technologies
37
145Presented at the Spring 2009 ICC Education Session Copyright GTRC 2009
CDFIDielectric Withstand
Josh Perkel
CDFI Research 146Presented at the Spring 2009 ICC Education Session Copyright GTRC 2009
Dielectric Withstand• Withstand techniques are most widely used diagnostic in
the USA.
• Most utilities use VLF (either sine or cosine-rectangular) in their withstand programs.
• Test duration and voltage are critical to performance on test and in service.
• Explored the concept of “Monitored” Withstand tests.
CDFI Research
147Presented at the Spring 2009 ICC Education Session Copyright GTRC 2009
Circuit Length [Conductor ft]
Perc
ent
840007200060000480003600024000120000
30
25
20
15
10
5
0
Median Length = 3500 ft
Length Distribution (Overall)
Wide variability in circuit lengths
CDFI Research 148Presented at the Spring 2009 ICC Education Session Copyright GTRC 2009
Length Effects• Comparison of withstand failure on test rates must include
length adjustments.
2000 ft.
500 ft. 500 ft. 500 ft.500 ft.
Censored
Failure
CDFI Research
38
149Presented at the Spring 2009 ICC Education Session Copyright GTRC 2009
Time on Test [Minutes]
Failu
res
on T
est
[%]
100.010.01.00.1
20
10
5
3
2
1
1000 Feet500 FeetNONE
AdjustmentLength
Utility I – Hybrid System
Time on Test [Minutes]
Failu
res
on T
est
[%]
100.010.01.00.1
20
10
5
3
2
1
1000 Feet500 FeetNONE
AdjustmentLength
Time on Test [Minutes]
Failu
res
on T
est
[%]
100.010.01.00.1
20
10
5
3
2
1
30
4.5%
2.4%
17.5%1000 Feet500 FeetNONE
AdjustmentLength
Performance at longer test times can be predicted.
Length Weighted Average FOT
30 Mins 2.7% 60 Mins 5.0%
CDFI Research 150Presented at the Spring 2009 ICC Education Session Copyright GTRC 2009
3.53.02.52.01.5
100
80
60
40
20
0
Test Voltage (U0 = Rated Voltage)
Surv
ivor
s [%
of
Test
ed]
IEEE Rec. Level
NEETRAC ExtrudedMixed (PILC and Extruded)ExtrudedPILC
Effect of Test Voltage
CDFI Research
151Presented at the Spring 2009 ICC Education Session Copyright GTRC 2009
VLF Lab Program
Josh Perkel
CDFI Research 152Presented at the Spring 2009 ICC Education Session Copyright GTRC 2009
152
Overview• Test program combining aging at U0 with multiple
applications of high voltage VLF.
• Uses field aged cable samples - one area within one utility.
• Evaluate the effects of – Voltage and time on the performance on test and – Subsequent reliability during service voltages.
Primary MetricSurvival during aging and testing
Secondary Metrics– Before and after each VLF application, PD at U0– Between Phase A & B IRC, PD (AC 2.2U0, DAC), Tan δ
CDFI Research
39
153Presented at the Spring 2009 ICC Education Session Copyright GTRC 2009
1: No Withstand
2: VLF2.2U015 Min
3: VLF3.6U0
120 Min
4: VLF2.5U060 Min
5: VLF2.2U0
120 Min
6: 60 Hz.3.6U0
0.25 Min
Withstand Testing Periods (variable durations)
Aging Periods
Phase A
End
CDFI Research
Failures are the primary metricfor evaluation
154Presented at the Spring 2009 ICC Education Session Copyright GTRC 2009
1: No Withstand
2: VLF2.2U015 Min
3: VLF3.6U0
120 Min
4: VLF2.5U060 Min
5: VLF2.2U0
120 Min
6: 60 Hz.3.6U0
0.25 Min
T1 T2 T3 T4
No Fails
14 Survive
No Fails14 Survive
3 VLF Fails11 Survive
2 VLF Fails12 Survive
No Fails14 Survive
2 60 Hz. Fails12 Survive
CDFI Research
No aging failures for any condition
155Presented at the Spring 2009 ICC Education Session Copyright GTRC 2009
Time of Failure on Test
Time on VFL Test (min)
Perc
ent
of 2
0 ft
Sam
ples
Fai
ling
100101
99
90
8070605040
30
20
10
5
3
2
1
15 30 60
63
253
C orrelation 0.997
Shape 1.46825Scale 254.187Failure 5C ensor 51
Max Time120 Mins
No Failures•Before 15 mins•After 60 mins
CDFI Research 156Presented at the Spring 2009 ICC Education Session Copyright GTRC 2009
Testing Voltage (Uo)
Perc
ent
7.06.05.04.03.02.01.5
99
90
8070605040
30
20
10
5
3
2
1
6.438
25.257
2.5
3.6
C orrelation 1.000
Shape 4.04729Scale 4.88340Failure 5C ensor 51
156
Voltage of Failure on Test
More failures occur at higher test voltages
CDFI Research
40
157Presented at the Spring 2009 ICC Education Session Copyright GTRC 2009
Failure Analyses - Trees & Defects in Cables
Distance Along Cable (ft)
Faile
d Sa
mpl
es
2520151050
D
C
B
A
DEFECTLARGE WATER TREEMEDIUM WATER TREESMALL WATER TREE
CDFI Research 158Presented at the Spring 2009 ICC Education Session Copyright GTRC 2009
158
VLF Test Program Summary• Analysis of Phase A is complete.
• Phase B (2U0 aging, 45°C Cosine Rectangular) underway.
• Phases A & B show that no VLF exposed samples have failed under 60 Hz aging @ Uo & 2Uo.
• Phase B tests showed two samples without VLF exposure failed during 60 Hz aging @ 2Uo.
• All failures occurred at the appropriate time. i.e. within the VLF testing periods.
• 80% (4 out of 5) of VLF failures between 15 and 60 mins
CDFI Research
159Presented at the Spring 2009 ICC Education Session Copyright GTRC 2009
Selecting a Diagnostic TechnologyKnowledge-Based System
Nigel Hampton
Selecting a Diagnostic Technology 160Presented at the Spring 2009 ICC Education Session Copyright GTRC 2009
KBS• Selecting the right diagnostic is not easy.
• No one diagnostic covers everything.
• How you measure is influenced by what you do with the results.
• The KBS captures the experience and knowledge of people who have been operating in the field
Selecting a Diagnostic Technology
41
161Presented at the Spring 2009 ICC Education Session Copyright GTRC 2009
Knowledge Based Systems• Knowledge-Based Systems are computer systems that are programmed to imitate human problem-solving.
• Uses a combination of artificial intelligence and reference to a database of knowledge on a particular subject.
• KBS are generally classified into:– Expert Systems– Case Based Reasoning– Fuzzy Logic Based Systems– Neural Networks
Selecting a Diagnostic Technology 162Presented at the Spring 2009 ICC Education Session Copyright GTRC 2009
Extruded Cable Diagnostics
Selecting a Diagnostic Technology
163Presented at the Spring 2009 ICC Education Session Copyright GTRC 2009
KBS Example
Selecting a Diagnostic Technology 164Presented at the Spring 2009 ICC Education Session Copyright GTRC 2009
Short Listing of Diagnostic Approaches
Exp
erts
Rec
omm
endi
ng a
Dia
gnos
tic
Tech
niqu
e (%
) 100
80
60
40
20
0
Simple Withstand Monitored WithstandDischarge
Simple Withstand Historical DataDischargeDielectric
Simple WithstandMonitored Withstand
EXPERTSBY FEWESTRECOMMENDED
BY MOST EXPERTSRECOMMENDED
Selecting a Diagnostic Technology
42
165Presented at the Spring 2009 ICC Education Session Copyright GTRC 2009
Impact of Remedial Action
• Hybrid Cable System• Most service failures occur in Accessories• Usual remediation is by replacement of cable sections
High
Low
Medium
Service Failure Rate
40 - 5025Paper
0 - 1042EPR
20 - 3033PE
Age[yrs]
Portion [%]
System Component
Selecting a Diagnostic Technology 166Presented at the Spring 2009 ICC Education Session Copyright GTRC 2009
Expe
rt R
ecom
men
dati
on (
%)
Replace Small Section
Replace Segment
Replace Accessories
Simple Withstand
Monitored Withstand
History & TDRDischarge
Dielectric
Simple Withstand
Monitored W ithstand
History & TDRDischarge
Dielectric
Simple Withstand
Monitored W ithstand
History & TDRDischarge
Dielectric
100
80
60
40
20
0
Hybrid Cable System
Expe
rt R
ecom
men
dati
on (
%)
Replace Small Section
Replace Segment
Replace Accessories
Simple Withstand
Monitored Withstand
History & TDRDischarge
Dielectric
Simple Withstand
Monitored W ithstand
History & TDRDischarge
Dielectric
Simple Withstand
Monitored W ithstand
History & TDRDischarge
Dielectric
100
80
60
40
20
0
Expe
rt R
ecom
men
dati
on (
%)
Replace Small Section
Replace Segment
Replace Accessories
Simple Withstand
Monitored Withstand
History & TDRDischarge
Dielectric
Simple Withstand
Monitored W ithstand
History & TDRDischarge
Dielectric
Simple Withstand
Monitored W ithstand
History & TDRDischarge
Dielectric
100
80
60
40
20
0
Selecting a Diagnostic Technology
Expe
rt R
ecom
men
dati
on (
%)
Replace Small Section
Replace Segment
Replace Accessories
Simple Withstand
Monitored Withstand
History & TDRDischarge
Dielectric
Simple Withstand
Monitored W ithstand
History & TDRDischarge
Dielectric
Simple Withstand
Monitored W ithstand
History & TDRDischarge
Dielectric
100
80
60
40
20
0
Most Recommended
167Presented at the Spring 2009 ICC Education Session Copyright GTRC 2009
Summary
Rick Hartlein
168Presented at the Spring 2009 ICC Education Session Copyright GTRC 2009
What we have learned about diagnostics (1)1. A developing database of field failure diagnostic data shows
that different diagnostic techniques can provide someindication about cable system condition.
2. Even if the diagnostics themselves are imprecise, diagnostic programs can be beneficial.
3. Benefits can be quantified, however this is not simple and requires effort.
4. Many different data analysis techniques, including some non conventional approaches, are needed to assess diagnostic effectiveness.
5. Utilities HAVE to act on ALL replacement/repair recommendations to get improved reliability.
Summary
43
169Presented at the Spring 2009 ICC Education Session Copyright GTRC 2009
What we have learned about diagnostics (2)6. PD, VLF, DC and Tan δ & VLF withstand tests detect
problems in the field and can be used to improve system reliability.
7. It is very difficult to predict whether or not the problems/defects detected by PD and Tan δ will lead to failure in the short/medium term.
8. PD assessments are good at establishing groups of cable system segments that are not likely to fail.
9. Tan δ measurements provide a number of interesting features for assessing the condition of cable systems.
10.Tan δ & PD measurements require interpretation to establish how to act.
Summary 170Presented at the Spring 2009 ICC Education Session Copyright GTRC 2009
11. Interpretation of PD measurements is more complex than interpretation of Tan δ measurements.
12. IRC & RV are particularly difficult to deploy in the field.
What we have learned about diagnostics (3)
Summary
171Presented at the Spring 2009 ICC Education Session Copyright GTRC 2009
Reflections• Approach to data analysis established in CDFI• Many questions answered, there still remain gaps in our
understanding of:– Benefits– Distinguishing anomalies from weaknesses
• Answers will come with continued analysis of field test data (diagnostic tests followed by circuit performance monitoring) as well as controlled laboratory tests.
• The potential value of continued analysis is high.
Summary