GTRI/DOE Disclaimer Cable Diagnostic Focused Initiative Regional ... CDFI Meeting.pdf · Cable...
Transcript of GTRI/DOE Disclaimer Cable Diagnostic Focused Initiative Regional ... CDFI Meeting.pdf · Cable...
1
1CDFI Meeting - Aug. 19-20 San Ramon, CA
Cable Diagnostic Focused InitiativeRegional Meeting
NEETRAC
Hosted byPacific Gas and Electric
San Ramon, CAAugust 19-20, 2009
2CDFI Meeting - Aug. 19-20 San Ramon, CA
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
3CDFI Meeting - Aug. 19-20 San Ramon, CA
PresentersDr. Nigel Hampton is the Program Manager for Reliability work at NEETRAC. He has worked in the Power Cable arena for more than 20 years. Nigel has a PhD in Physics from the University of Bath UK. He is currently the vice-chair of the Insulated Conductor Committee’s subcommittee on diagnostic testing (Subcommittee F).
Dr. Joshua Perkel is a Research Engineer in the Assessment group at NEETRAC. He has worked in the Power Cable arena for more than 5 years. Josh holds a PhD in electrical engineering from the Georgia Institute of Technology.
4CDFI Meeting - Aug. 19-20 San Ramon, CA
CDFI ContributorsNEETRAC
Rick Hartlein (PI)Thomas ParkerJoshua Perkel
Jorge AltamiranoTim AndrewsYamille del ValleNigel Hampton (Co-PI)
Georgia Tech - ECEMiroslav BegovicRon HarleyJ.C. HernandezSalman Mohagheghi
IREQJean-Francois Drapeau
2
5CDFI Meeting - Aug. 19-20 San Ramon, CA
Day 1
Lunch12:00 – 13:00
SAGE Concept14:30 – 14:45
Diagnostic Testing Technologies (Part I)16:30 – 17:00
Break14:45 – 15:00
Diagnostic Accuracies16:00 – 16:30Case Study: Roswell15:00 – 16:00
Cable System Failure Process14:00 – 14:30CDFI Background/Overview 13:30 – 14:00
NEETRAC Overview13:10 – 13:30Welcome13:00 – 13:10
TopicTime
6CDFI Meeting - Aug. 19-20 San Ramon, CA
Day 2
Continental Breakfast07:30 – 08:00
Lunch12:00 – 13:00
Accuracies Really Matter09:30 – 10:00
Selecting a Diagnostic Testing Technology11:25 – 11:45The Things We Know Now That We Did Not Know Before10:15 – 11:20
Review Day 108:00 – 08:15
Summary11:45 – 12:00
Break10:00 – 10:15
Diagnostic Testing Technologies (Part II)08:15 – 09:30
TopicTime
7CDFI Meeting - Aug. 19-20 San Ramon, CA
Day 1
8CDFI Meeting - Aug. 19-20 San Ramon, CA
Outline• NEETRAC Overview• 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
3
9CDFI Meeting - Aug. 19-20 San Ramon, CA
NEETRAC Overview
10CDFI Meeting - Aug. 19-20 San Ramon, CA
Background
• Created in 1996 when Georgia Power donated the facilities of its Research Center to Georgia Tech.
• Set up as a self supporting center within the School of Electrical and Computer Engineering of the Georgia Tech.
• NEETRAC is a membership based center, conducting research programs for the Electric Energy Transmission and Distribution Industry.
NEETRAC Overview
11CDFI Meeting - Aug. 19-20 San Ramon, CA
NEETRAC Mission & VisionMissionTo provide a venue where NEETRAC Staff, NEETRAC Members and the Georgia Tech Academic community can collaborate to solve problems in the T&D Arena.
VisionWe will build on our expertise to become the leading national Center for collaborative applied and strategic research and development for electric transmission and distribution.
NEETRAC Overview 12CDFI Meeting - Aug. 19-20 San Ramon, CA
Members 2009-20101. 3M2. ABB3. Ameren Services4. American Electric Power5. Baltimore Gas & Electric6. British Columbia Hydro7. Borealis Compounds LLC8. Con Edison9. Cooper Power Systems10. Dominion/Virginia Power11. Dow Chemical Company12. Duke Energy13. Entergy14. Exelon15. First Energy16. Florida Power & Light17. GRESCO Utility Supply
18. Hubbell19. NRECA20. NSTAR21. PacifiCorp22. Prysmian Cables & Systems 23. Public Service Electric & Gas24. S&C Electric Company25. South Carolina Electric & Gas26. Southern California Edison27. Southern Company28. Southern States29. Southwire30. Thomas and Betts/Homac31. TVA32. tyco / Raychem33. Zenergy Power
NEETRAC Overview
4
13CDFI Meeting - Aug. 19-20 San Ramon, CA
NEETRAC Membership Growth
20102008200620042002200019981996
40
30
20
10
0
Year
Mem
bers
2009
NEETRAC Overview 14CDFI Meeting - Aug. 19-20 San Ramon, CA
Members• Utility Members
– Provide > 50% of power sold in the US– Serve over 64,000,000 customers
• Manufacturing Members– Primary suppliers of T&D equipment to electric
utilities in the United States
NEETRAC Overview
15CDFI Meeting - Aug. 19-20 San Ramon, CA
Focus Areas Developed
Training/Education
Safety
Power Quality/Grounding
Operation, Installation, Design
System Analysis
Forensics
Condition Assessment
Asset Management
Reliability
System Enhancements
Research
New Product DevelopmentNew Technology/Research
Equipment Spec. & Test Protocol Development
Engineering Analysis & Support
Product Evaluation
Application Research
Hardware/Equipment Testing
FOCUS SEGMENTSPRIMARY FOCUS AREA
NEETRAC Overview 16CDFI Meeting - Aug. 19-20 San Ramon, CA
Facilities: High Voltage Lab
NEETRAC Overview
5
17CDFI Meeting - Aug. 19-20 San Ramon, CA
Facilities: Low Voltage & Mechanical Lab
NEETRAC Overview 18CDFI Meeting - Aug. 19-20 San Ramon, CA
Ploss
Investment
ΔV
γ r
aγ
Direction of Drop Movement
γ r
aγ
Direction of Drop Movement
NEETRAC Overview
19CDFI Meeting - Aug. 19-20 San Ramon, CA
Tensile and Impact Tests on 12-ft Grounding JumpersC-clamp on 2-in pin to Flat-face Clamp on 5/8-in pin, Tension versus Clamp Displacement
-500
0
500
1000
1500
2000
2500
3000
0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4Actuator Displacement (inches)
Tens
ion
(lb)
Sample J4 Impact
Sample J4 Tensile
Sample J5, Impact
Sample J5 Tensile
Sample I4, Impact
Sample I5, Impact
Sample I6, Impact
NEETRAC Overview 20CDFI Meeting - Aug. 19-20 San Ramon, CA
Staff• 25 Research Staff
– Ph.D degees (EE & Physics)– M.S. degrees (EE, IE, & ME)– Bachelors degrees (EE & ME)
• 5 Administrative and IT Support
• 1 Coop Students
NEETRAC Overview
6
21CDFI Meeting - Aug. 19-20 San Ramon, CA
This is NEETRAC• 25 Research Staff• 5 Administrative
Support Staff• Academic Faculty• Co-op and Graduate
Students
NEETRAC Overview 22CDFI Meeting - Aug. 19-20 San Ramon, CA
Outline• NEETRAC Overview• 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
23CDFI Meeting - Aug. 19-20 San Ramon, CA
CDFI Background
24CDFI Meeting - Aug. 19-20 San Ramon, CA
• 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
7
25CDFI Meeting - Aug. 19-20 San Ramon, CA
Composition of US MV systemIn
stal
led
Capa
city
(%
)
UNKNOWNTRXLPEEPRXLPEHMWPEPILC
100
90
80
70
60
50
40
30
20
10
0
25
50
75
CDFI Background/Overview 26CDFI Meeting - Aug. 19-20 San Ramon, CA
Failure Split
Unknown1.1%Terminations
5.6%
Splices37.1% Cable
56.2%
CDFI Background/Overview
27CDFI Meeting - Aug. 19-20 San Ramon, CA
• 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 28CDFI Meeting - Aug. 19-20 San Ramon, CA
NEETRAC Members
Non NEETRACMembers Supporters
Dept of Energy
Diagnostic Providers
CDFI
CDFI Background/Overview
8
29CDFI Meeting - Aug. 19-20 San Ramon, CA
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 30CDFI Meeting - Aug. 19-20 San Ramon, CA
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
31CDFI Meeting - Aug. 19-20 San Ramon, CA
CDFI Activities
CDFI
Analysis Lab Studies
Field Studies Dissemination
CDFI Background/Overview 32CDFI Meeting - Aug. 19-20 San Ramon, CA
CDFI ActivitiesCDFI
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
9
33CDFI Meeting - Aug. 19-20 San Ramon, CA
CDFI ActivitiesLab
Studies
VLF Withstand Tan δ PD
Test TimeTest VoltageForensics
Time StabilityVoltage Stability
Non-Uniform DegradationNeutral Corrosion
CalibrationPhase Pattern
Feature ExtractionClassification
CDFI Background/Overview 34CDFI Meeting - Aug. 19-20 San Ramon, CA
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
35CDFI Meeting - Aug. 19-20 San Ramon, CA
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 36CDFI Meeting - Aug. 19-20 San Ramon, CA
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
10
37CDFI Meeting - Aug. 19-20 San Ramon, CA
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 38CDFI Meeting - Aug. 19-20 San Ramon, CA
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
39CDFI Meeting - Aug. 19-20 San Ramon, CA
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
40CDFI Meeting - Aug. 19-20 San Ramon, CA
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
11
41CDFI Meeting - Aug. 19-20 San Ramon, CA
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 42CDFI Meeting - Aug. 19-20 San Ramon, CA
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)
43CDFI Meeting - Aug. 19-20 San Ramon, CA
• 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 44CDFI Meeting - Aug. 19-20 San Ramon, CA
QUESTIONS
12
45CDFI Meeting - Aug. 19-20 San Ramon, CA
Outline• NEETRAC Overview• 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
46CDFI Meeting - Aug. 19-20 San Ramon, CA
How things fail and what fails have a big impact on the selection of diagnostics
Cable System Failure Process
47CDFI Meeting - Aug. 19-20 San Ramon, CA
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 48CDFI Meeting - Aug. 19-20 San Ramon, CA
Failure Rates
100
80
60
40
20
0
Failu
re R
ate
[#/1
00 M
iles/
Yea
r]
Lower Quartile: 1.6Median: 3.5Upper Quartile: 8 Mean: 12Max: 140
Peak at 140
Cable System Failure Process
13
49CDFI Meeting - Aug. 19-20 San Ramon, CA
Failure Rate Estimates – By Equipment
Unk RateTerm RateSplice RateCable Rate
8.0
7.0
6.0
5.0
4.0
3.0
2.5
2.0
1.5
1.0
0.5
0.0
Failu
re R
ate
- M
onte
Car
lo E
stim
ate
(#/1
00m
iles/
yr)
Cable System Failure Process 50CDFI Meeting - Aug. 19-20 San Ramon, CA
Major Cable Components
Jacket (Recommended)
Metallic Shield/Neutral
Insulation Shield
Insulation
Conductor
Cable System Failure Processhttp://www.otds.co.uk/cables.php
Conductor Shield
Extruded PILC
51CDFI Meeting - Aug. 19-20 San Ramon, CA
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 52CDFI Meeting - Aug. 19-20 San Ramon, CA
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 growthElectrical tree growing
from water treeCable System Failure Process
14
53CDFI Meeting - Aug. 19-20 San Ramon, CA
Defect Types in Extruded Cable Accessories
Cable System Failure Process 54CDFI Meeting - Aug. 19-20 San Ramon, CA
Diagnostics used in Challenging Areas
Cable System Failure Process
55CDFI Meeting - Aug. 19-20 San Ramon, CA
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 56CDFI Meeting - Aug. 19-20 San Ramon, CA
QUESTIONS
15
57CDFI Meeting - Aug. 19-20 San Ramon, CA
Outline• NEETRAC Overview• 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
58CDFI Meeting - Aug. 19-20 San Ramon, CA
SAGE Approach to
Diagnostic Programs
59CDFI Meeting - Aug. 19-20 San Ramon, CA
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 60
CDFI Meeting - Aug. 19-20 San Ramon, CA
SAGE at WorkFailures [#]
Time
Selection
Action
Generation
Evaluation
Decreasing Failures
Increasing Failures
SAGE Concept
16
61CDFI Meeting - Aug. 19-20 San Ramon, CA
Failures [#]
Time
Decreasing Failures
Increasing Failures
Continued Failure Increase
SAGE Concept 62CDFI Meeting - Aug. 19-20 San Ramon, CA
When to deploy diagnostics
Time (Years)
Cabl
e Sy
stem
Per
form
ance
403020100
Operational Stress
Condition AssessmentCommissioning
SAGE Concept
63CDFI Meeting - Aug. 19-20 San Ramon, CA
Global ContextComparison with many tests
DatabasesStandards
Context – is important
Local ContextComparisons within one area
DataGeneration from
Diagnostic Measurement
SAGE Concept 64CDFI Meeting - Aug. 19-20 San Ramon, CA
QUESTIONS
17
65CDFI Meeting - Aug. 19-20 San Ramon, CA
Outline• NEETRAC Overview• 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
66CDFI Meeting - Aug. 19-20 San Ramon, CA
Case StudyRoswell, GA
November 2008 & January 2009
TDRTan Delta
Monitored WithstandOffline PD
67CDFI Meeting - Aug. 19-20 San Ramon, CA
67
Roswell Map
Case Study: Roswell 68CDFI Meeting - Aug. 19-20 San Ramon, CA
SELECTION
Case Study: Roswell
18
69CDFI Meeting - Aug. 19-20 San Ramon, CA
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 70CDFI Meeting - Aug. 19-20 San Ramon, CA
Knowledge Based Selection System
Case Study: Roswell
71CDFI Meeting - Aug. 19-20 San Ramon, CA
KBS Demo
Case Study: Roswell 72CDFI Meeting - Aug. 19-20 San Ramon, CA
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
19
73CDFI Meeting - Aug. 19-20 San Ramon, CA
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 74CDFI Meeting - Aug. 19-20 San Ramon, CA
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
Case Study: Roswell
75CDFI Meeting - Aug. 19-20 San Ramon, CA
ACTION
Case Study: Roswell 76CDFI Meeting - Aug. 19-20 San Ramon, CA
Initial Corrective Action Options
• Replace splices only – no detailed records assume 12 splices.
• Complete system replacement.
Case Study: Roswell
20
77CDFI Meeting - Aug. 19-20 San Ramon, CA
GENERATION
Case Study: Roswell 78CDFI Meeting - Aug. 19-20 San Ramon, CA
Overhead and Cabinet Terminations
Case Study: Roswell
79CDFI Meeting - Aug. 19-20 San Ramon, CA
Tan δ Monitored Withstand
Case Study: Roswell 80CDFI Meeting - Aug. 19-20 San Ramon, CA
If this had been a Simple Withstand
Length Tested (miles)1086420
18 Segments Tested
No Failures On Test
Case Study: Roswell
21
81CDFI Meeting - Aug. 19-20 San Ramon, CA
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
82CDFI Meeting - Aug. 19-20 San Ramon, CA
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
83CDFI Meeting - Aug. 19-20 San Ramon, CA
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 84CDFI Meeting - Aug. 19-20 San Ramon, CA
Targeted Offline PD (VLF)
Case Study: Roswell
22
85CDFI Meeting - Aug. 19-20 San Ramon, CA
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 86CDFI Meeting - Aug. 19-20 San Ramon, CA
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
87CDFI Meeting - Aug. 19-20 San Ramon, CA
EVALUATION
Case Study: Roswell 88CDFI Meeting - Aug. 19-20 San Ramon, CA
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 in 7 months since
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) in 8 months since test
Case Study: Roswell
23
89CDFI Meeting - Aug. 19-20 San Ramon, CA
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 90CDFI Meeting - Aug. 19-20 San Ramon, CA
QUESTIONS
91CDFI Meeting - Aug. 19-20 San Ramon, CA
Break
92CDFI Meeting - Aug. 19-20 San Ramon, CA
Outline• NEETRAC Overview• 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
24
93CDFI Meeting - Aug. 19-20 San Ramon, CA
Diagnostic Accuracies
94CDFI Meeting - Aug. 19-20 San Ramon, CA
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
95CDFI Meeting - Aug. 19-20 San Ramon, CA
Diagnostic Measurements and Failures• Symptoms are difficult to relate to future failures unless they are
in the extremes.
Pro
babi
lity
“Good” “Bad”?
Diagnostic Measurement
No Failure Failure
Diagnostic Accuracies 96CDFI Meeting - Aug. 19-20 San Ramon, CA
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
25
97CDFI Meeting - Aug. 19-20 San Ramon, CA
Diagnostic OperationApplying the diagnostic will separate the population into:• No Action Required group• Action Required group
But if the diagnostic is imperfect...
No Action Required Action Required
Diagnostic Accuracies 98CDFI Meeting - Aug. 19-20 San Ramon, CA
Complimentary Diagnoses
Online PD Offline PD
VLF TDNo ActionActionNo Test
Category
33.8%
52.1%
14.1%
69.0%
11.3%
19.7%
83.1%
4.2%
12.7%
VLF TD 25% Offline PD 36% Online PD 79%Ratio Action / No Action
3 Service Failures
since testing completedDiagnostic Accuracies
99CDFI Meeting - Aug. 19-20 San Ramon, CA
CDFI Accuracies
Action
Agr
eem
ent
ActDont_Act
Agree
Disagree
Agreement btw Online Offline * Online PDAgreement btw Online VLF TD * Online PDAgreement btw Offline VLF TD * Offline PD
Variable
CDFI diagnostic accuracies are based on service performance (failures) not diagnostic agreement.
Diagnostic Accuracies 100CDFI Meeting - Aug. 19-20 San Ramon, CA
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
26
101CDFI Meeting - Aug. 19-20 San Ramon, CA
“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
102CDFI Meeting - Aug. 19-20 San Ramon, CA
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 [%
]
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 [%
]
Diagnostic Accuracies
103CDFI Meeting - Aug. 19-20 San Ramon, CA
QUESTIONS
104CDFI Meeting - Aug. 19-20 San Ramon, CA
Outline• NEETRAC Overview• 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
27
105CDFI Meeting - Aug. 19-20 San Ramon, CA
Diagnostic Testing Technologies
106CDFI Meeting - Aug. 19-20 San Ramon, CA
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
107CDFI Meeting - Aug. 19-20 San Ramon, CA
Utility Use of Diagnostics
Diagnostic Testing Technologies 108CDFI Meeting - Aug. 19-20 San Ramon, CA
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
28
109CDFI Meeting - Aug. 19-20 San Ramon, CA
No TestingTesting - one techniqueTesting - > one technique
27.8%
30.6%
41.7%
Survey of Use of Diagnostics
Diagnostic Testing Technologies 110CDFI Meeting - Aug. 19-20 San Ramon, CA
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
111CDFI Meeting - Aug. 19-20 San Ramon, CA
Technologies• Simple Dielectric Withstand• Dielectric Loss (Tan δ & Dielectric Spectroscopy)• Time Domain Reflectometry (TDR)• Online Partial Discharge (PD) • Offline Partial Discharge (PD)• Isothermal Relaxation Current (IRC)• Recovery Voltage (RV)• Combined Diagnostics
Diagnostic Testing Technologies 112CDFI Meeting - Aug. 19-20 San Ramon, CA
Global ContextComparison with many tests
DatabasesStandards
Context
Local ContextComparisons within one area
DataGeneration from
Diagnostic Measurement
Diagnostic Testing Technologies
29
113CDFI Meeting - Aug. 19-20 San Ramon, CA
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 114CDFI Meeting - Aug. 19-20 San Ramon, CA
Simple Dielectric Withstand
115CDFI Meeting - Aug. 19-20 San Ramon, CA
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.
Simple Dielectric Withstand 116CDFI Meeting - Aug. 19-20 San Ramon, CA
Withstand Test Process
HOLDEARLY
Time
Voltage
t = 0 tTest
Voltages and Times for VLF covered in IEEE Std. 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
Hold Entry
Ramp Entry
Simple Dielectric Withstand
30
117CDFI Meeting - Aug. 19-20 San Ramon, CA
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
Simple Dielectric Withstand 118CDFI Meeting - Aug. 19-20 San Ramon, CA
DataGeneration from
Diagnostic Measurement
Simple Dielectric Withstand
119CDFI Meeting - Aug. 19-20 San Ramon, CA
Test Sequences
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
Simple Dielectric Withstand 120CDFI Meeting - Aug. 19-20 San Ramon, CA
Local ContextComparisons within one area
Simple Dielectric Withstand
31
121CDFI Meeting - Aug. 19-20 San Ramon, CA
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.
Simple Dielectric Withstand 122CDFI Meeting - Aug. 19-20 San Ramon, CA
“Early” Phase Matters
Time on Test [Minutes]
Failu
res
on T
est
[% o
f Te
sted
]
10.001.000.100.01
60
50
40
30
20
10
Early Hold
60 % of failures on test occurred during
“Early” phase
Simple Dielectric Withstand
123CDFI Meeting - Aug. 19-20 San Ramon, CA
Time on Test (Mins)
Failu
re s
On
Test
- F
OT
(% o
f Se
ctio
ns T
este
d)
151050
0.5
0.4
0.3
0.2
0.1
0.0
STA
RT
OF
HO
LD P
HA
SE
151050
STA
RT
OF
HO
LD P
HA
SEDC VLF
1327
VoltageFeeder
0.11%
0.16%
“Early” and “Hold” Phases
Simple Dielectric Withstand
Length Adjusted
Difference between VLF and DC is primarily result of “Early” phase
124CDFI Meeting - Aug. 19-20 San Ramon, CA
“Early” Phase – Ramp Entry Example
Voltage [U0]
Failu
res
on T
est
[% o
f To
tal T
ests
]
10.01.00.1
80
70
60
50
40
30
20
10
In this case, 60 % of the tests produced a failure before reaching the target test voltage.
Simple Dielectric Withstand
32
125CDFI Meeting - Aug. 19-20 San Ramon, CA
“Early” and “Hold” Failure Mechanisms (VLF)
Time on Test [Minutes]
Failu
res
on T
est
[% o
f To
tal T
este
d]
100.050.010.05.01.00.50.1
20
15
10
5
1
“Early”Phase
“Hold” Phase
Simple Dielectric Withstand 126CDFI Meeting - Aug. 19-20 San Ramon, CA
“Early” Phase – Hold Entry
Time on Test [Minutes]
Failu
res
on T
est
[% o
f To
tal T
este
d]
10.05.01.00.50.1
5
1
Simple Dielectric Withstand
“Early” phase accounts for 30 % of failures on test.
127CDFI Meeting - Aug. 19-20 San Ramon, CA
Global ContextComparison with many tests
DatabasesStandards
Simple Dielectric Withstand 128CDFI Meeting - Aug. 19-20 San Ramon, CA
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
9700 Conductor Miles>2000 Conductor Miles
0.3 Conductor Miles
Simple Dielectric Withstand
33
129CDFI Meeting - Aug. 19-20 San Ramon, CA
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.
Simple Dielectric Withstand 130CDFI Meeting - Aug. 19-20 San Ramon, CA
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
Day
s63
7 Da
ys
10%
1247
Day
s
2247
Day
s
15 Min @ 2.5U030 Min @ 1.8U0
2650 Conductor Miles
224763730 Min @ 1.8 U0
472
Time to Failure5%
[Days]
124715 Min @ 2.5 U0
Time to Failure10%
[Days]Test Conditions
Simple Dielectric Withstand
131CDFI Meeting - Aug. 19-20 San Ramon, CA
Performance After Test – Pass/No Pass
Simple Dielectric Withstand
100010010
20
10
5
3
2
1
0.1
T ime to Failure [Days]
Fa
ilu
res
on
Te
st [
% o
f T
ota
l T
est
ed
]
1
133
215
PassNo Pass - Repaired
Initial Test Result
100010010
20
10
5
3
2
1
0.1
T ime to Failure [Days]
Fa
ilu
res
on
Te
st [
% o
f T
ota
l T
est
ed
]
1
36 1228
PassNo Pass - Repaired
Initial Test Result
15 Minute 30 MinuteNot Length Adjusted
132CDFI Meeting - Aug. 19-20 San Ramon, CA
Difference between Passing and not Passing
Time to Failure for 1 % of Tested Segments[Days]
1228363021513315
No Pass - RepairedPass
Test Duration[Min]
Segments that fail on test and subsequently repaired perform better in service.
34
133CDFI Meeting - Aug. 19-20 San Ramon, CA
What does this mean for Withstand?• The technique is widely used by utilities
• Tested circuits display improved reliability
• Circuits normally Pass the tests
• Multiple / cascading failures are rare
• IEEE400.2 recommended times (30 mins) and voltages seem to give good service performance
134CDFI Meeting - Aug. 19-20 San Ramon, CA
What does this mean for Withstand?• IEEE400.2 recommended times (30 mins) and voltages
seem to give good service performance
• Modifications to IEEE400.2 recommendations need to be considered very carefully
• Voltage & test time cannot be determined independently
• Many test fails occur early in the test, useful information is revealed by tracking of these times / voltages of failure
• More failures on test does not mean fewer service fails
135CDFI Meeting - Aug. 19-20 San Ramon, CA
QUESTIONS
136CDFI Meeting - Aug. 19-20 San Ramon, CA
Day 2
35
137CDFI Meeting - Aug. 19-20 San Ramon, CA
Outline• NEETRAC Overview• 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
138CDFI Meeting - Aug. 19-20 San Ramon, CA
Dielectric Loss (Tan δ)
139CDFI Meeting - Aug. 19-20 San Ramon, CA
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.• Step voltage up to pre determined level with post test analysis
Tan δ 140CDFI Meeting - Aug. 19-20 San Ramon, CA
Dielectric Loss (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 the equivalent circuit consists of two parameters (IEEE
Std. 400):• Resistor• Capacitor
• When voltage is applied to the cable, the total current is the sum of the capacitor current and resistor current.
Tan δ
36
141CDFI Meeting - Aug. 19-20 San Ramon, CA
Cable System Equivalent
T C S C TCable system (cable, splices, and
terminations) is reduced to simple circuit.
Tan δ 142CDFI Meeting - Aug. 19-20 San Ramon, CA
DataGeneration from
Diagnostic Measurement
Tan δ
143CDFI Meeting - Aug. 19-20 San Ramon, CA
Tan δ Ramp 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
Tan δ 144CDFI Meeting - Aug. 19-20 San Ramon, CA
Local ContextComparisons within one area
Tan δ
37
145CDFI Meeting - Aug. 19-20 San Ramon, CA
Tan δ Data for EPR Cable Systems
Voltage (kV)
Tan
Del
ta (
1e-3
)
1211109876543
25
20
15
10
5 ConcernLowest
ConcernHighest
Tan δ 146CDFI Meeting - Aug. 19-20 San Ramon, CA
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
Tan δ
147CDFI Meeting - Aug. 19-20 San Ramon, CA
Lengths within a Local Region
Length (ft)
Tan
Del
ta (
1e-3
)
50004000300020001500
10
1
Tan δ 148CDFI Meeting - Aug. 19-20 San Ramon, CA
Global ContextComparison with many tests
DatabasesStandards
Tan δ
38
149CDFI Meeting - Aug. 19-20 San Ramon, CA
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
Tan δ 150CDFI Meeting - Aug. 19-20 San Ramon, CA
Tan δ Interpretation
Tip Up
Tan
Del
ta
-1010-1-
-
150
6
0
No Action
Further Study
Action Required
Based on 258 Conductor Miles
Tan δ
151CDFI Meeting - Aug. 19-20 San Ramon, CA
Tan δ Correlation with VLF Withstand
Length (ft)
Tan
D (
1e-3
)
1000100
1000.0
100.0
10.0
1.0
0.11000100
?Filled
Unfilled
Basic Type
Fail Subsequent VLF Withstand Pass Subsequent VLF Withstand
Tan δ 152CDFI Meeting - Aug. 19-20 San Ramon, CA
Elasped Time between test and failure in service at May 09 (Month)
Perc
ent
101FOT
40
30
20
10
5
3
2
1
ARFSNA
Action
Tan δ Performance Curves
Elasped Time between test and failure in service at May 09 (Month)
Perc
ent
101FOT
40
30
20
10
5
3
2
1
ACTION REQUIREDFURTHER STUDYNO ACTION
Action
Elasped Time between test and failure in service at May 09 (Month)
Perc
ent
101FOT
40
30
20
10
5
3
2
1
ACTION REQUIREDFURTHER STUDYNO ACTION
Action
Elasped Time between test and failure in service at May 09 (Month)
Perc
ent
101FOT
40
30
20
10
5
3
2
1
ACTION REQUIREDFURTHER STUDYNO ACTION
Action
Elasped Time between test and failure in service at May 09 (Month)
Perc
ent
101FOT
40
30
20
10
5
3
2
1
293236
91011
1.71.92.3
14
4
0.66012 24
ACTION REQUIREDFURTHER STUDYNO ACTION
Action
Tan δ
39
153CDFI Meeting - Aug. 19-20 San Ramon, CA
What does this mean for Tan δ?
• Provides information on the whole cable system• Most useful features are
– Time Stability– Differential Tan δ (Tip Up)
• Higher loss correlates with increased probability of failure• Comparisons provide very useful information
– Length effects– Adjacent sections / phases
• Existing levels in IEEE Std. 400 are too conservative. Newer (higher) levels to be in IEEE Std. 400.2 revision
154CDFI Meeting - Aug. 19-20 San Ramon, CA
Time Domain Reflectometry
155CDFI Meeting - Aug. 19-20 San Ramon, CA
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.
TDR 156CDFI Meeting - Aug. 19-20 San Ramon, CA
TDR Principles
Near End
TDREquipment
Far End
JointL
Joint
TDR
40
157CDFI Meeting - Aug. 19-20 San Ramon, CA
Wet Joint
• Feeder had two splice failures just before the test.
• Water ingress was detected with the TDR.
• Failure on 01/17/2008 at detected water ingress location.
• Water ingress confirmed by tests and repair crew.
Water ingress location as seen by the TDR
TDR 158CDFI Meeting - Aug. 19-20 San Ramon, CA
TDR Field Measurements
500040003000200010000
A-1
B-2
C-3
Distance from Cubicle 2 [ft]
Phas
e
circles)Anomalous TDR reflections (open
TDR
159CDFI Meeting - Aug. 19-20 San Ramon, CA
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
Measurements made with TDR
TDR 160CDFI Meeting - Aug. 19-20 San Ramon, CA
What does this mean for TDR?
• All diagnostics rely on the neutral, TDR helps to establish its condition.
• Length and accessory information are very important in establishing the context of diagnostic findings.
• Unusual TDR traces can diagnose unusual features in their own right.
41
161CDFI Meeting - Aug. 19-20 San Ramon, CA
Online Partial Discharge
162CDFI Meeting - Aug. 19-20 San Ramon, CA
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 performed by a service provider.• Different implementations of the overall approach• Assessment criteria are unique to each embodiment of the
technologyOnline PD
163CDFI Meeting - Aug. 19-20 San Ramon, CA
DataGeneration from
Diagnostic Measurement
Online PD 164CDFI Meeting - Aug. 19-20 San Ramon, CA
Discharge Occurrence
No PD PD
Online PD
42
165CDFI Meeting - Aug. 19-20 San Ramon, CA
Local ContextComparisons within one area
Online PD 166CDFI Meeting - Aug. 19-20 San Ramon, CA
Distribution of PD along Lengths• 5000 ft. portion of sample feeder
• Mixture of different PD levels for different sections and accessories.
Cable Section Accessory
No PDPD
Online PD
167CDFI Meeting - Aug. 19-20 San Ramon, CA
Global ContextComparison with many tests
DatabasesStandards
Online PD 168CDFI Meeting - Aug. 19-20 San Ramon, CA
Where is PD found?
Accessory54.0%
Cable46.0%
Online PD
43
169CDFI Meeting - Aug. 19-20 San Ramon, CA
Variability in PD LocationPe
rcen
tage
of
PD (
%)
AccessoryCable
100
80
60
40
20
0
Online PD 170CDFI Meeting - Aug. 19-20 San Ramon, CA
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%
226 Conductor Miles
Online PD
171CDFI Meeting - Aug. 19-20 San Ramon, CA
Level Based Reporting Systems• Level-based (i.e. “1, 2, 3” , “Defer, Repair, Replace”, “Act , Don’t Act”
etc.) reporting systems are increasingly common.
• Level systems, on their own, can have limited meaning for utilities.
• Levels clearly indicate a hierarchy– “5” worse than “4” “Replace” worse than “Defer”
• No sense of the magnitude of the difference– How much worse is “Act” than “Don’t Act” in terms of service
performance?
• Comparisons / interpretation of different level-based reporting systems is difficult.
Need to associate meaning with the levels
Level Based Reporting 172CDFI Meeting - Aug. 19-20 San Ramon, CA
Online PD Performance Curve
Time to Failure (Years)
Perc
ent
20.010.05.01.00.50.1
99
90807060504030
20
10
5
3
2
1
3%
18%
89%
2
345
Level
Level Based Reporting
44
173CDFI Meeting - Aug. 19-20 San Ramon, CA
Alternate Interpretation
89518433
< 32<< 31
Alternate Class(based on probability of failure)
Original Level
Class 18 has 6 times poorer endurance than Class 3
Class 89 is 5 times poorer than Class 18
Level Based Reporting 174CDFI Meeting - Aug. 19-20 San Ramon, CA
Probabilistic Approach – Online PD
Days Between Test & Failure
Perc
ent
1000100101
99
9080706050403020
10
532
1
0.01
No PDPD
PD Class
Level Based Reporting
175CDFI Meeting - Aug. 19-20 San Ramon, CA
Variability in Diagnostic Results
Perc
ent
Level 5Level 4Level 3Level 2Level 1
90
80
70
60
50
40
30
20
10
0
Level 5Level 4Level 3Level 2Level 1
Accessory Cable
Online PD 176CDFI Meeting - Aug. 19-20 San Ramon, CA
How often is PD found?
PD O
ccur
ence
(#
/100
0 ft
)
All PDAccessory PDCable PD
0.35
0.30
0.25
0.20
0.15
0.10
0.05
0.00
1 PD signal every 4000 ft4.3 signals / 1000 ft
Online PD
45
177CDFI Meeting - Aug. 19-20 San Ramon, CA
Estimated Failure Reduction
Time since Start [Days]
Cum
mul
ativ
e Fa
ilure
s [#
]
1000100
100
10
11000100
ACCESSORY CABLE
Levels 4/5 ReplacedAll Segments Left in Service
14 Avoided Fails45 Actions for
23 Avoided Fails52 Actions for
Online PD 178CDFI Meeting - Aug. 19-20 San Ramon, CA
What does this mean for Online PD?
• Highly degraded systems most easily differentiated• Not necessarily easy to deploy – sensor placement and
manhole access can be challenging• Signal analysis is labor intensive• Data for level interpretation is available• Trending is likely to be valuable, incorporating this in a
level-based reporting system can be a challenge• Baseline (when new) studies likely to be valuable• Active failure mechanisms need to involve discharges• Can localize to accessory and cable segments
179CDFI Meeting - Aug. 19-20 San Ramon, CA
Offline Partial Discharge
180CDFI Meeting - Aug. 19-20 San Ramon, CA
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
• Step voltage up to pre determined level with post test analysisOffline PD
46
181CDFI Meeting - Aug. 19-20 San Ramon, CA
DataGeneration from
Diagnostic Measurement
Offline PD 182CDFI Meeting - Aug. 19-20 San Ramon, CA
PD Pulse
140 mV
180 pC
Offline PD
183CDFI Meeting - Aug. 19-20 San Ramon, CA
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]
Offline PD 184CDFI Meeting - Aug. 19-20 San Ramon, CA
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
Offline PD
47
185CDFI Meeting - Aug. 19-20 San Ramon, CA
Local ContextComparisons within one area
Offline PD 186CDFI Meeting - Aug. 19-20 San Ramon, CA
PD Charge Magnitude Distributions
Apparent Charge Magnitude [pC]
Perc
ent
6005004003002001000
20
15
10
5
0
Apparent Charge Magnitude [pC]
Perc
ent
6005004003002001000
20
15
10
5
0
XLPE
Offline PD
187CDFI Meeting - Aug. 19-20 San Ramon, CA
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
Offline PD 188CDFI Meeting - Aug. 19-20 San Ramon, CA
Global ContextComparison with many tests
DatabasesStandards
Offline PD
48
189CDFI Meeting - Aug. 19-20 San Ramon, CA
189
Location of PD
Termination26.3%
Splice34.3%
Cable39.4%
60.6% of PD sites detected in accessories
222 Conductor Miles
Offline PD 190CDFI Meeting - Aug. 19-20 San Ramon, CA
Offline PD Test Sequence• Testing sequence for 16,000 ft.
No PD
PDOffline PD
191CDFI Meeting - Aug. 19-20 San Ramon, CA
PD Location
Location [% of Circuit Length]
Perc
ent
9075604530150
18
16
14
12
10
8
6
4
2
0
Terminations
Cable & Splices
Offline PD 192CDFI Meeting - Aug. 19-20 San Ramon, CA
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
Offline PD
49
193CDFI Meeting - Aug. 19-20 San Ramon, CA
What does this mean for Offline PD?
• Highly degraded systems most easily differentiated• Signal analysis can be labor intensive• Data for level interpretation could be available• Trending is likely to be very valuable• Incorporating trending in a level-based reporting system
can be a challenge• Baseline (when new) studies likely to be very valuable• Active failure mechanisms need to involve discharges• Can localize to accessory and within short cable length
within a segment
194CDFI Meeting - Aug. 19-20 San Ramon, CA
Isothermal Relaxation CurrentRecovery Voltage
195CDFI Meeting - Aug. 19-20 San Ramon, CA
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.
IRC 196CDFI Meeting - Aug. 19-20 San Ramon, CA
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.
Recovery Voltage
50
197CDFI Meeting - Aug. 19-20 San Ramon, CA
What does this mean for IRC & RV?
• Use limited to evaluation studies in the laboratory• Possibly too sensitive for field use
198CDFI Meeting - Aug. 19-20 San Ramon, CA
Combined Diagnostics
Multiple degradation mechanisms mean that two diagnostics are often better than one
Combined Diagnostics
199CDFI Meeting - Aug. 19-20 San Ramon, CA
No TestingTesting - one techniqueTesting - > one technique
27.8%
30.6%
41.7%
Survey of Use of Diagnostics
Combined Diagnostics 200CDFI Meeting - Aug. 19-20 San Ramon, CA
Multiple Diagnostics
Tan Delta / PDVLF / Tan Delta
Category
75.0%
25.0%
Combined Diagnostics
51
201CDFI Meeting - Aug. 19-20 San Ramon, CA
What Diagnostics are Combined
Local
DC Withstand
PD VLF Withstand
DC Leakage
Tan δ
Glo
bal
TDR
Combined Diagnostics 202CDFI Meeting - Aug. 19-20 San Ramon, CA
Drawbacks of a Single Approach• Each diagnostic looks for symptoms of one failure
mechanism– Voids and water trees cannot generally be detected by
a single technique
• Overlooks short term time evolution of diagnostic measurements
• Technique specific:– Withstand – No idea by how much segment passed– Tan δ – Cannot detect voids or electrical trees– PD – Cannot detect water trees (water filled voids)
Combined Diagnostics
203CDFI Meeting - Aug. 19-20 San Ramon, CA
BAD
GOOD
Advantage of Multiple Diagnostics
Bad
Diagnostic 1
Goo
d
Good Bad?
?
Diagnostic 2
204CDFI Meeting - Aug. 19-20 San Ramon, CA
DataGeneration from
Diagnostic Measurement
Combined Diagnostics
52
205CDFI Meeting - Aug. 19-20 San Ramon, CA
Tan δ Ramp
Time [min]
Tan-
delt
a [1
e-3]
54321
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
Combined Diagnostics 206CDFI Meeting - Aug. 19-20 San Ramon, CA
Tan δ Monitored Withstand
Time [min]
Tan-
delt
a [1
e-3]
1512.5107.552.50
1.8
1.7
1.6
1.5
1.4
1.3
1.2
1.1
1.0
through the 15 minute withstandStability of Tan-delta monitored
Combined Diagnostics
207CDFI Meeting - Aug. 19-20 San Ramon, CA
Tan δ Ramp & Monitored Withstand
Time [min]
Tan-
delt
a [1
e-3]
76543210
180
160
140
120
100
80
60
40
20
0.51.01.51.7
[p.u.]Voltage
FailureSegment HL_23_22Elbow Failure Hampton Leas
Combined Diagnostics 208CDFI Meeting - Aug. 19-20 San Ramon, CA
After Repair…
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
Failure
After Failure
Segment HL_23_22Elbow Failure Hampton Leas
Combined Diagnostics
53
209CDFI Meeting - Aug. 19-20 San Ramon, CA
Global ContextComparison with many tests
DatabasesStandards
Combined Diagnostics 210CDFI Meeting - Aug. 19-20 San Ramon, CA
Tan δ 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
Combined Diagnostics
211CDFI Meeting - Aug. 19-20 San Ramon, CA
QUESTIONS
212CDFI Meeting - Aug. 19-20 San Ramon, CA
Outline• NEETRAC Overview• 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
54
213CDFI Meeting - Aug. 19-20 San Ramon, CA
Accuracies Revisited
Why do they matter?
214CDFI Meeting - Aug. 19-20 San Ramon, CA
Consequence
Diagnostic Program Costs Cost [$]
Selection
Diagnostic
CorrectiveActions Total Diagnostic
Program Cost
Accuracies Really Matter
215CDFI Meeting - Aug. 19-20 San Ramon, CA
Recall the Example...No Action Required Action Required
Avoided Corrective Actions
Avoided service failures
Accuracies Really Matter 216CDFI Meeting - Aug. 19-20 San Ramon, CA
Incorrect Diagnosis
Future service failures Unneeded Corrective Actions
No Action Required Action Required
Accuracies Really Matter
55
217CDFI Meeting - Aug. 19-20 San Ramon, CA
Benefit and LossCost [$]
Selection
CorrectiveActions
Diagnostic
ConsequenceAlternate
Program 1
AlternateProgram 2
BENEFIT
LOSS
Accuracies Really Matter 218CDFI Meeting - Aug. 19-20 San Ramon, CA
Considerations
• Diagnostic program economic calculations are based on ability to predict future failures.
• Total diagnostic program cost is more sensitive to certain elements than others.– Failure Rate– Diagnostic Accuracy– Failure Consequence
Accuracies Really Matter
219CDFI Meeting - Aug. 19-20 San Ramon, CA
Uncertainty in Diagnostic Program Costs Cost [$]
Diagnostic
Selection
Consequence
CorrectiveActions
Program Cost
Range
Accuracies Really Matter 220CDFI Meeting - Aug. 19-20 San Ramon, CA
AlternateProgram 1
BENEFIT
LOSS
Program Cost
Range
Cost [$]Uncertainty in Diagnostic Program Costs
Accuracies Really Matter
56
221CDFI Meeting - Aug. 19-20 San Ramon, CA
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 222CDFI Meeting - Aug. 19-20 San Ramon, CA
Failures Over Time
No Action Required Action Required
Year12345
Accuracies Really Matter
223CDFI Meeting - Aug. 19-20 San Ramon, CA
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 224CDFI Meeting - Aug. 19-20 San Ramon, CA
Elasped Time between test and failure in service at May 09 (Month)
Perc
ent
101FOT
40
30
20
10
5
3
2
1
ARFSNA
Action
Probabilistic Approach - Tan δ
Elasped Time between test and failure in service at May 09 (Month)
Perc
ent
101FOT
40
30
20
10
5
3
2
1
ACTION REQUIREDFURTHER STUDYNO ACTION
Action
Elasped Time between test and failure in service at May 09 (Month)
Perc
ent
101FOT
40
30
20
10
5
3
2
1
ACTION REQUIREDFURTHER STUDYNO ACTION
Action
Elasped Time between test and failure in service at May 09 (Month)
Perc
ent
101FOT
40
30
20
10
5
3
2
1
ACTION REQUIREDFURTHER STUDYNO ACTION
Action
Elasped Time between test and failure in service at May 09 (Month)
Perc
ent
101FOT
40
30
20
10
5
3
2
1
293236
91011
1.71.92.3
14
4
0.66012 24
ACTION REQUIREDFURTHER STUDYNO ACTION
Action
Accuracies Really Matter
57
225CDFI Meeting - Aug. 19-20 San Ramon, CA
QUESTIONS
226CDFI Meeting - Aug. 19-20 San Ramon, CA
Outline• NEETRAC Overview• 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
227CDFI Meeting - Aug. 19-20 San Ramon, CA
The Things We Know Now That We Did Not Know Before
228CDFI Meeting - Aug. 19-20 San Ramon, CA
By Diagnostic Technique
VLF DC Tan Delta
PD On PD Off TDR
IRC DAC
No UseOccasionalStandardTesting
Category
CDFI Research
58
229CDFI Meeting - Aug. 19-20 San Ramon, CA
CDFI Work in Lab and Field
• Dielectric Withstand– Simple– VLF Laboratory Study
• Dielectric Loss• VLF Tan δ
– Monitored Withstand• Partial Discharge• Offline 60 Hz.
CDFI Research 230CDFI Meeting - Aug. 19-20 San Ramon, CA
CDFIDielectric Withstand
Dielectric Withstand
231CDFI Meeting - Aug. 19-20 San Ramon, CA
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.
Dielectric Withstand 232CDFI Meeting - Aug. 19-20 San Ramon, CA
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
Dielectric Withstand
59
233CDFI Meeting - Aug. 19-20 San Ramon, CA
Length Adjustments• Comparison of withstand failure on test rates must include
length adjustments.
2000 ft.
Choose an appropriate base length
Dielectric Withstand 234CDFI Meeting - Aug. 19-20 San Ramon, CA
Length Adjustments• Comparison of withstand failure on test rates must include
length adjustments.
2000 ft.
500 ft. 500 ft. 500 ft.500 ft.
Censored
Failure
Dielectric Withstand
235CDFI Meeting - Aug. 19-20 San Ramon, CA
Length Adjustments
• Base length must be a meaningful length (50 ft is probably not a useful length).
• Two sets of censored segments:– Pass Segments - All segments censored at test
duration– No Pass Segments
• 1 failed segment• remaining segments censored at failure time
• Multiple failure modes must be dealt with appropriately.
Dielectric Withstand 236CDFI Meeting - Aug. 19-20 San Ramon, CA
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%
Dielectric Withstand
60
237CDFI Meeting - Aug. 19-20 San Ramon, CA
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
Dielectric Withstand 238CDFI Meeting - Aug. 19-20 San Ramon, CA
VLF Lab Program
Dielectric Withstand
239CDFI Meeting - Aug. 19-20 San Ramon, CA
239
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
Dielectric Withstand 240CDFI Meeting - Aug. 20-21 San Ramon, CA
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
Failures are the primary metricfor evaluation
61
241CDFI Meeting - Aug. 20-21 San Ramon, CA
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 242CDFI Meeting - Aug. 20-21 San Ramon, CA
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 Failures
No Failures
3 VLF FailuresNo Aging Failures
2 VLF FailuresNo Aging Failures
No Failures
2 60 Hz. FailuresNo Aging Failures
243CDFI Meeting - Aug. 20-21 San Ramon, CA
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 Failures
No Failures
3 VLF FailuresNo Aging Failures
2 VLF FailuresNo Aging Failures
No Failures
2 60 Hz. FailuresNo Aging Failures
244CDFI Meeting - Aug. 19-20 San Ramon, CA
Failures on Test
100101
90
50
10
1100101
90
50
10
1
100101
90
50
10
1100101
90
50
10
1
3Uo Rated, Sine
Time of VLF Application (mins)
Prob
abili
ty o
f Fa
ilure
(%
)
15 60
10
71
2.5Uo Rated, Sine15 60
10
134
3Uo Rated, Cosine15 60
10
8
2.5Uo Rated, Cosine15 60
10
63
Uo, Ambient - Phase I & II
2Uo, 45C - Phase III
Dielectric Withstand
62
245CDFI Meeting - Aug. 19-20 San Ramon, CA
Voltage Effect on Times to Failure
3.02.82.62.42.2
180
160
140
120
100
80
60
40
20
03.02.82.62.42.2
Phase I & II - Uo / RT ageing, Sine
Test Voltage [Uo]
Tim
e to
10%
Fai
lure
[m
ins]
Phase III - 2Uo / 45C ageing, Cosine
Both curves show that higher voltage leads to increased failure rate
Dielectric Withstand 246CDFI Meeting - Aug. 19-20 San Ramon, CA
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
Dielectric Withstand
247CDFI Meeting - Aug. 19-20 San Ramon, CA
247
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 @ U0 & 2U0.
• Phase B tests shows two samples without VLF exposure failed during 60 Hz aging @ 2U0.
• VLF failures on test:– Less than 15 mins: 12 % (2 failures)– 15 – 60 mins: 71 % (12 failures)
Dielectric Withstand 248CDFI Meeting - Aug. 19-20 San Ramon, CA
CDFIDielectric Loss
Tan δ
63
249CDFI Meeting - Aug. 19-20 San Ramon, CA
Prevailing View – Tan Delta
Tan δ
Tip Up[2U0 – 1U0]
Importance
Tan δ 250CDFI Meeting - Aug. 19-20 San Ramon, CA
CDFI Suggestion – Tan Delta
Tan δTime
Stability
Tip Up[1.5U0 – 0.5U0]
Tan δ[U0]
Importance
Tan δ
251CDFI Meeting - Aug. 19-20 San Ramon, CA
Tan δ Time Stability
Tan-delta stdv. @ 0.1 Hz (1.5 Uo) Diagnostic Rank
Brea
kdow
n Pe
rfor
man
ce R
ank
76543210
7
6
5
4
3
2
1
0
0.160.0
[Hz]FrequencyBreakdown
Tan δ 252CDFI Meeting - Aug. 19-20 San Ramon, CA
VLF Tan Delta of Cable Systems
Tan Delta at Uo (E-3)
Perc
ent
1000.0100.010.01.00.1
99.9
99
90
8070605040
30
20
10
5
3
2
1
FilledPaperPE
Ins Class
•>650 segments•Mean Length 2000ft•Total length >250 conductor miles
Can segregated based on areas where the curves breakDefine areas that are “normal” and “unusual”
Tan δ
64
253CDFI Meeting - Aug. 19-20 San Ramon, CA
Cable System – Global Assessment
Tip Up (1e-3)
Tan
Del
ta (
1e-3
)
-1010-1-
-
150
6
0
Unfilled Polyolefin Insulations
No Action
Further Study
Action Required
Tan δ 254CDFI Meeting - Aug. 19-20 San Ramon, CA
Service Performance / Accuracy
Elasped Time between test and failure in service at May 09 (Month)
Perc
ent
101FOT
40
30
20
10
5
3
2
1
293236
91011
1.71.92.3
14
4
0.66012 24
ACTION REQUIREDFURTHER STUDYNO ACTION
Action
Tan δ
255CDFI Meeting - Aug. 19-20 San Ramon, CA
CDFIPartial Discharge
PD 256CDFI Meeting - Aug. 19-20 San Ramon, CA
CDFI Work
• Analysis of historical PD field test data
• Classification
• Characterization of field samples by PD measurement in laboratory.
• Feature Extraction for Classification
PD
65
257CDFI Meeting - Aug. 19-20 San Ramon, CA
PD Charge Magnitude Distributions
540450360270180900-90
40
30
20
10
0
Charge Magnitude [pC]
Perc
ent
Cable_Failed_PCCable_NOFailed_PC
Variable
540450360270180900-90
40
30
20
10
0
Charge Magnitude [pC]
Perc
ent
Cable_Failed_PCCable_NOFailed_PC
Variable
540450360270180900-90
40
30
20
10
0
Charge Magnitude [pC]
Perc
ent
Cable_Failed_PCCable_NOFailed_PC
Variable
PD 258CDFI Meeting - Aug. 19-20 San Ramon, CA
PD Inception Voltage
2.72.42.11.81.51.20.9
30
25
20
15
10
5
0
Inception Voltage [U0]
Perc
ent
Cable_Failed_IVCable_NOFailed_IV
Variable
2.72.42.11.81.51.20.9
30
25
20
15
10
5
0
Inception Voltage [U0]
Perc
ent
Cable_Failed_IVCable_NOFailed_IV
Variable
2.72.42.11.81.51.20.9
30
25
20
15
10
5
0
Inception Voltage [U0]
Perc
ent
Cable_Failed_IVCable_NOFailed_IV
Variable
PD
259CDFI Meeting - Aug. 19-20 San Ramon, CA
BAD
GOOD
Multi Feature Classification
Bad
Criterion 1
Goo
d
Good Bad?
?
Criterion 2
PD 260CDFI Meeting - Aug. 19-20 San Ramon, CA
Classification - PD Magnitude & PDIV
Neighbors Used in Classification [#]
Succ
es R
ate
[% o
f Te
sted
]
3937353331292725232119171513119753
100
90
80
70
60
50
40
30
20
10
0
fail_successnofail_successOverall_success
Variable
Neighbors Used in Classification [#]
Succ
es R
ate
[% o
f Te
sted
]
3937353331292725232119171513119753
100
90
80
70
60
50
40
30
20
10
0
fail_successnofail_successOverall_success
Variable
pC and PDIV are not sufficient to get high classification accuracy
PD
66
261CDFI Meeting - Aug. 19-20 San Ramon, CA
Cluster No. Feature Name1 Pos. Phase Range [deg]2 Pos. Mean Phase [deg]
3
Pos. Qmax [pC]Neg. Qmax [pC]Neg. Qmean [pC]Pos. Qmean [pC]Pos. Mean Energy [pC*V]Pos. Max Energy [pC*V]Neg. Max Energy [pC*V]
Neg. Mean Energy [pC*V]
4 Neg. Phase Range [deg]5 Neg. Mean Phase [deg]
6 DMean Energy Ratio
7 Nw [pulses/cycle]
Cluster No. Feature Name1 Pos. Phase Range [deg]2 Pos. Mean Phase [deg]
3
Pos. Qmax [pC]Neg. Qmax [pC]Neg. Qmean [pC]Pos. Qmean [pC]Pos. Mean Energy [pC*V]Pos. Max Energy [pC*V]Neg. Max Energy [pC*V]
Neg. Mean Energy [pC*V]
4 Neg. Phase Range [deg]5 Neg. Mean Phase [deg]
6 DMean Energy Ratio
7 Nw [pulses/cycle]
PD Lab Data - Cluster Variable Analysis
261PD 262CDFI Meeting - Aug. 19-20 San Ramon, CA
Partial Discharge Diagnostic Features
Sim
ilari
ty L
evel
[%
]
Nw [p
ulses
/cycl
e]
Mean En
ergy
Rati
oD
Neg.
Mean Ph
ase [d
eg]
Neg.
Phas
e Ran
ge [d
eg]
Neg. M
ean E
nergy [
pC*k
V]
Neg.
Max En
ergy
[pC*
kV]
Pos.
Max En
ergy
[pC*
kV]
Pos.
Mean E
nergy [
pC*k
V]
Pos.
Qmea
n [pC
]
Neg. Q
mea
n [pC
]
Neg.
Qmax [p
C]
Pos.
Qmax
[pC]
Pos.
Mean Ph
ase [d
eg]
Pos.
Phas
e Ran
ge [d
eg]
15.18
43.45
71.73
100.00
Partial Discharge Diagnostic Features
Sim
ilari
ty L
evel
[%
]
Nw [p
ulses
/cycl
e]
Mean En
ergy
Rati
oD
Neg.
Mean Ph
ase [d
eg]
Neg.
Phas
e Ran
ge [d
eg]
Neg. M
ean E
nergy [
pC*k
V]
Neg.
Max En
ergy
[pC*
kV]
Pos.
Max En
ergy
[pC*
kV]
Pos.
Mean E
nergy [
pC*k
V]
Pos.
Qmea
n [pC
]
Neg. Q
mea
n [pC
]
Neg.
Qmax [p
C]
Pos.
Qmax
[pC]
Pos.
Mean Ph
ase [d
eg]
Pos.
Phas
e Ran
ge [d
eg]
15.18
43.45
71.73
100.00
PD Lab Data - Cluster Variable Analysis
262Partial Discharge Diagnostic Features
Sim
ilari
ty L
evel
[%
]
Nw [p
ulses
/cycl
e]
Mean En
ergy
Rati
oD
Neg.
Mean Ph
ase [d
eg]
Neg.
Phas
e Ran
ge [d
eg]
Neg. M
ean E
nergy [
pC*k
V]
Neg.
Max En
ergy
[pC*
kV]
Pos.
Max En
ergy
[pC*
kV]
Pos.
Mean E
nergy [
pC*k
V]
Pos.
Qmea
n [pC
]
Neg. Q
mea
n [pC
]
Neg.
Qmax [p
C]
Pos.
Qmax
[pC]
Pos.
Mean Ph
ase [d
eg]
Pos.
Phas
e Ran
ge [d
eg]
15.18
43.45
71.73
100.00
Partial Discharge Diagnostic Features
Sim
ilari
ty L
evel
[%
]
Nw [p
ulses
/cycl
e]
Mean En
ergy
Rati
oD
Neg.
Mean Ph
ase [d
eg]
Neg.
Phas
e Ran
ge [d
eg]
Neg. M
ean E
nergy [
pC*k
V]
Neg.
Max En
ergy
[pC*
kV]
Pos.
Max En
ergy
[pC*
kV]
Pos.
Mean E
nergy [
pC*k
V]
Pos.
Qmea
n [pC
]
Neg. Q
mea
n [pC
]
Neg.
Qmax [p
C]
Pos.
Qmax
[pC]
Pos.
Mean Ph
ase [d
eg]
Pos.
Phas
e Ran
ge [d
eg]
15.18
43.45
71.73
100.00
Partial Discharge Diagnostic Features
Sim
ilari
ty L
evel
[%
]
Nw [p
ulses
/cycl
e]
Mean En
ergy
Rati
oD
Neg.
Mean Ph
ase [d
eg]
Neg.
Phas
e Ran
ge [d
eg]
Neg. M
ean E
nergy [
pC*k
V]
Neg.
Max En
ergy
[pC*
kV]
Pos.
Max En
ergy
[pC*
kV]
Pos.
Mean E
nergy [
pC*k
V]
Pos.
Qmea
n [pC
]
Neg. Q
mea
n [pC
]
Neg.
Qmax [p
C]
Pos.
Qmax
[pC]
Pos.
Mean Ph
ase [d
eg]
Pos.
Phas
e Ran
ge [d
eg]
15.18
43.45
71.73
100.00
50 % Similarity Level
1 2 3 4 5 6 7
3a 3b
PD
263CDFI Meeting - Aug. 19-20 San Ramon, CA
QUESTIONS
264CDFI Meeting - Aug. 19-20 San Ramon, CA
Outline• NEETRAC Overview• 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
67
265CDFI Meeting - Aug. 19-20 San Ramon, CA
Selecting a Diagnostic TechnologyKnowledge-Based System
Selecting a Diagnostic Technology 266CDFI Meeting - Aug. 19-20 San Ramon, CA
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
267CDFI Meeting - Aug. 19-20 San Ramon, CA
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 268CDFI Meeting - Aug. 19-20 San Ramon, CA
Extruded Cable Diagnostics
Selecting a Diagnostic Technology
68
269CDFI Meeting - Aug. 19-20 San Ramon, CA
KBS Example
Selecting a Diagnostic Technology 270CDFI Meeting - Aug. 19-20 San Ramon, CA
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
271CDFI Meeting - Aug. 19-20 San Ramon, CA
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 272CDFI Meeting - Aug. 19-20 San Ramon, CA
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
69
273CDFI Meeting - Aug. 19-20 San Ramon, CA
QUESTIONS
274CDFI Meeting - Aug. 19-20 San Ramon, CA
Outline• NEETRAC Overview• 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
275CDFI Meeting - Aug. 19-20 San Ramon, CA
Summary
276CDFI Meeting - Aug. 19-20 San Ramon, CA
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
70
277CDFI Meeting - Aug. 19-20 San Ramon, CA
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 difficult to predict whether or not the problems/defects detected by PD or Tan δ will lead to failure.
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 278CDFI Meeting - Aug. 19-20 San Ramon, CA
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
279CDFI Meeting - Aug. 19-20 San Ramon, CA
DC
Voltage
Time
Diagnostic Information
Info
rmat
ion
Con
tent
High
Low
Ease of Utility ImplementationSimple Some Skill Difficult
VLFDAC
Combined Diagnostics 280CDFI Meeting - Aug. 19-20 San Ramon, CA
Voltage
Time
Info
rmat
ion
Con
tent
High
Low
Ease of Utility ImplementationSimple Some Skill Difficult
VLFDAC
TDRamp
PDRamp
Diagnostic Information
Combined Diagnostics
DC
71
281CDFI Meeting - Aug. 19-20 San Ramon, CA
DC
Voltage
Time
Info
rmat
ion
Con
tent
High
Low
Ease of Utility ImplementationSimple Some Skill Difficult
VLFDAC
TDRamp
PDRamp
Diagnostic Information
Combined Diagnostics
Preset Test ProtocolsAnalysis of data performed after
tests are completed.
Adaptive Test ProtocolsTests adjusted in real time according to
analysis/decisions made during each test.
282CDFI Meeting - Aug. 19-20 San Ramon, CA
Voltage
Time
Info
rmat
ion
Con
tent
High
Low
Simple Some Skill Difficult
VLF
DC
DAC
TDRamp
PDRamp
TD/PD Monitored Withstand
Ease of Utility Implementation
Diagnostic Information
Combined Diagnostics
?
283CDFI Meeting - Aug. 19-20 San Ramon, CA
Voltage
Time
Info
rmat
ion
Con
tent
High
Low
Simple Some Skill Difficult
VLFDAC
TDRamp
PDRamp
TD Ramp w/TD Monitored Withstand
PD Ramp w/PD Monitored Withstand
Ease of Utility Implementation
Diagnostic Information
Combined Diagnostics
TD/PD Monitored Withstand
?
DC
284CDFI Meeting - Aug. 19-20 San Ramon, CA
Info
rmat
ion
Con
tent
High
Low
Simple Some Skill Difficult
VLFDAC
TDRamp
PDRamp
TD & PD Ramp TD & PD Monitored Withstand
Ease of Utility Implementation
Voltage
Time
Diagnostic Information
Combined Diagnostics
TD Ramp w/TD Monitored Withstand
PD Ramp w/PD Monitored Withstand
TD/PD Monitored Withstand
?
DC
72
285CDFI Meeting - Aug. 19-20 San Ramon, CA
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 286CDFI Meeting - Aug. 19-20 San Ramon, CA
CDFI Phase 1 Extension
Schedule: October 1, 2009 - September 30, 2010
Tasks• VLF Withstand• Defects• Field Surveys• Regional Meetings
Summary
287CDFI Meeting - Aug. 19-20 San Ramon, CA
CDFI Phase IISchedule: January 2010? (3 Year Duration)
Tasks• High Voltage Testing • Commissioning Tests• Field Demonstrator• Field Testing
– Revisits/Trending– Challenging Utility Regions
• Diagnostic Reference Handbook• Knowledge-Based System
Summary
Phase II ParticipantsCurrent CDFI
EPRIDOE
New entrants
288CDFI Meeting - Aug. 19-20 San Ramon, CA
QUESTIONS