Cable Diagnostic Focused Initiative Regional Meeting NEETRAC

Transcription

1 Cable Diagnostic Focused Initiative Regional Meeting NEETRAC Hosted by Consolidated Edison Company of New York New York, NY October 28-29,

2 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-FC02-04CH

3 Presenters Dr. Nigel Hampton is the Program Manager for Reliability work at NEETRAC. He has worked in the Power Cable arena for more than 20 years. Vice-chair of the ICC subcommittee on diagnostic testing (Sub F) Convenor of CIGRE WGB1.28 on On-site Partial Discharge Assessment of HV and EHV cable systems. 3 3

4 Mr. Rick Hartlein is the Director of NEETRAC and Principal Investigator for this project. He has over 30 years of experience performing research projects on Power Cable Systems. He actively participates in the development of industry standards and specifications for underground cable systems and has served as Chair of ICC. Presenters 4 4

5 Presenters 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. 5 5

6 CDFI Contributors NEETRAC Jorge Altamirano Nigel Hampton (Co-PI) Tim Andrews Rick Hartlein (PI) Yamille del Valle Thomas Parker Bryan Davant Joshua Perkel Stacy Elledge Dean Williams Barry Fairley Georgia Tech - ECE Miroslav Begovic Ron Harley J.C. Hernandez Salman Mohagheghi IREQ Jean-Francois Drapeau 6 6

7 Day 1 Time 12:00 13:00 13:00 13:10 13:10 13:30 13:30 14:00 14:00 14:30 14:30 14:40 14:40 15:10 15:10 15:25 15:25 16:20 16:20 17:15 18:00 Topic Lunch Welcome NEETRAC Overview CDFI Background/Overview Cable System Failure Process SAGE Concept Analytical Techniques & Failure Rates Break Diagnostic Testing Technologies Case Study: Roswell Dinner 7 7

8 Day 2 Time 07:30 08:00 08:00 08:15 08:15 08:45 08:45 09:30 09:30 10:00 10:00 10:15 10:15 11:30 11:30 12:00 12:00 13:00 Topic Continental Breakfast Review Day 1 Diagnostic Accuracies Accuracies Really Matter The Things That Are Much Clearer Now - Research Break The Things That Are Much Clearer Now - Research Summary Lunch 8 8

9 Outline NEETRAC Overview CDFI Background/Overview Cable System Failure Process SAGE Concept Analytical Techniques & Failure Rates Diagnostic Testing Technologies Case Study: Roswell Diagnostic Accuracies Accuracies Really Matter The Things That Are Much Clearer Now CDFI Research Summary 9 9

10 NEETRAC Overview 10 10

11 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 11 11

12 Mission NEETRAC Mission & Vision To provide a venue where NEETRAC Staff, NEETRAC Members and the Georgia Tech Academic community can collaborate to solve problems in the T&D Arena. Vision We 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 12 12

13 Members Utility Members Serve over 70,000,000 customers Manufacturing Members Primary suppliers of T&D equipment to electric utilities in the United States NEETRAC Overview 13 13

14 NEETRAC Membership Growth Members Year NEETRAC Overview 14 14

15 Members M 2. ABB 3. Ameren Services 4. American Electric Power 5. Baltimore Gas & Electric 6. British Columbia Hydro 7. Borealis Compounds LLC 8. Con Edison 9. Cooper Power Systems 10. Dominion/Virginia Power 11. Dow Chemical Company 12. Duke Energy 13. Entergy 14. Exelon 15. First Energy 16. Florida Power & Light 17. GRESCO Utility Supply 18. Hubbell 19. NRECA 20. NSTAR 21. PacifiCorp 22. Prysmian Cables & Systems 23. Public Service Electric & Gas 24. S&C Electric Company 25. South Carolina Electric & Gas 26. Southern California Edison 27. Southern Company 28. Southern States 29. Southwire 30. Thomas and Betts/Homac 31. TVA 32. tyco / Raychem 33. Zenergy Power NEETRAC Overview 15 15

16 Focus Areas Developed PRIMARY FOCUS AREA FOCUS SEGMENTS Application Research Product Evaluation Hardware/Equipment Testing Engineering Analysis & Support Equipment Spec. & Test Protocol Development New Product Development New Technology/Research Research System Enhancements Asset Management Reliability Condition Assessment Forensics Operation, Installation, Design Power Quality/Grounding System Analysis Safety Training/Education NEETRAC Overview 16 16

17 Investment ΔV P loss NEETRAC Overview 17 17

18 200 Observed Failures Failure Estimate 150 Failures [#/Year] Historical Prediction Year NEETRAC Overview 18 18

19 Outline NEETRAC Overview CDFI Background/Overview Cable System Failure Process SAGE Concept Analytical Techniques & Failure Rates Diagnostic Testing Technologies Case Study: Roswell Diagnostic Accuracies Accuracies Really Matter The Things That Are Much Clearer Now CDFI Research Summary 19 19

20 CDFI Background 20 20

21 Why do we need diagnostics? 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. Cable Failures per Year 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. CDFI Background/Overview 21 21

22 Where has CDFI focused? Element Voltage Level Test Type Cable Diagnostics Data Lab Studies CDFI Focus, Phase I MV Condition Assessment Service Aged Currently in use in US Utility Distribution System Field Aged Cable 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. CDFI Background/Overview 22 22

23 Where has CDFI focused? Element Voltage Level Test Type Cable Diagnostics Data Lab Studies CDFI Focus, Phase I MV Condition Assessment Service Aged Currently in use in US Utility Distribution System Field Aged Cable Not Included in CDFI, Phase I HV Commissioning Laboratory Aged Not used in US Industrial & Transmission Accessories 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. CDFI Background/Overview 23 23

24 Diagnostic Providers NEETRAC Members CDFI Dept of Energy Non NEETRAC Members Supporters CDFI Background/Overview 24 24

25 Participants American Electric Power Ameren Cablewise / Utilx CenterPoint Energy Con Edison Cooper Power Systems Duke Power Company Exelon (Commonwealth Edison & PECO) First Energy Florida Power & Light Georgia Tech GRESCO HDW Electronics High Voltage, Inc HV Diagnostics HV Technologies Hydro Quebec IMCORP NRECA PacifiCorp (added mid 2005) Pacific Gas & Electric (added Jan 06) PEPCO Oncor (TXU) Prysmian Public Service Electric & Gas Tyco / Raychem Southern California Edison Southern Company Southwire CDFI Background/Overview 25 25

26 CDFI Activities CDFI Analysis Lab Studies Field Studies Dissemination CDFI Background/Overview 26 26

27 CDFI Activities Lab Studies (Service Aged Cables) VLF Withstand Tan δ PD Test Time Test Voltage Forensics Time Stability Voltage Stability Non-Uniform Degradation Neutral Corrosion Calibration Phase Pattern Feature Extraction Classification CDFI Background/Overview 27 27

28 CDFI Activities Field Studies Georgia Power XLPE Jkt & UnJkt 21 Conductor Miles Duke XLPE & Paper Jkt & UnJkt 29 Conductor Miles Offline PD (0.1Hz) Offline PD (60Hz) Tan δ Monitored Withstand Offline PD (0.1Hz) Tan δ Monitored Withstand Evans Macon Roswell Charlotte * 2 Cincinnati Clemson Morresville CDFI Background/Overview 28 28

29 CDFI Activities Utility Data Con Ed Com Ed PPL Alabama Power Keyspan DC Withstand Online PD VLF Withstand Offline PD (60Hz) Online PD Tan Delta VLF Withstand Offline PD (0.1Hz) Tan Delta Online PD Offline PD (0.1Hz) Tan Delta CDFI Background/Overview 29 29

30 CDFI Activities Utility Data FPL PEPCO PG&E ONCOR Ameren Offline PD (60Hz) VLF Withstand Offline PD (60Hz) Offline PD (0.1Hz) Online PD VLF Withstand Offline PD (60Hz) Online PD Tan δ Offline PD (60Hz) Online PD Offline PD (60Hz) CDFI Background/Overview 30 30

31 Dataset Sizes Data Type Technique Laboratory [Conductor miles] Field [Conductor miles] DC Withstand - 78,105 Monitored Withstand PD Offline Diagnostic PD Online Tan δ VLF Withstand 1.5 9,810 IRC Service Performance ALL 89,000 CDFI Background/Overview 31 31

32 Data Perspective Results presented must be viewed in light: CDFI focus Available data The data you will see here are Real Generated by or provided to utilities Not as complete as we would like CDFI Background/Overview 32 32

33 Benefits from Diagnostic Programs Decreasing failures associated with diagnostics and actions Log Cumulative Failures Program Initiated 20 failures 900 failures 250 failures Time [Days] 3000 CDFI Background/Overview 33 33

34 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 34 34

35 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 35 35

36 Where we are today (2) 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 36 36

37 QUESTIONS 37 37

38 Outline NEETRAC Overview CDFI Background/Overview Cable System Failure Process SAGE Concept Analytical Techniques & Failure Rates Diagnostic Testing Technologies Case Study: Roswell Diagnostic Accuracies Accuracies Really Matter The Things That Are Much Clearer Now CDFI Research Summary 38 38

39 How things fail and what fails have a big impact on the selection of diagnostics Cable System Failure Process 39 39

40 Cable System Defects Several causes exist for defects to occur in cable systems. Manufacturing, installation, aging, etc. The characteristics of a defect affect the influence it has on the system s performance. Defects represent non-uniform regions in the insulation material these lead to stress enhancement. Defect Stress Enhancement? Failure Cable System Failure Process 40 40

41 Extruded Cable Defects Oxidized Insulation Water trees Hot Conductor Crack Void Contaminant Corroded Neutral Cable System Failure Process 41 41

42 Number per ft Panel variable: Sample Contaminants in Cables s vintage Unjacketed XLPE Variable Contaminants Shield Defects These 1970 s cables have 1 defect / 1 ft 42 defects / 20 ft 9 defects / 20 ft Today s cables have 1 defect / 23 ft 0 defects / 20 ft 23 defects / 20 ft Length (ft) Cable System Failure Process 42 42

43 Defect Types in Extruded Cable Accessories Cable System Failure Process 43 43

44 Treeing Degrades Insulation Materials Treeing weakens the cable system does not necessarily mean that failure is imminent Two basic types they are fundamentally different beasts Water Tree Bowtie Electrical Vented Electrical Vented Bow tie Concern Treeing is a complicated phenomenon. Cable System Failure Process 44 44

45 Conversion of Water to Electrical Trees Electrical tree growing from water tree Acts as a stress enhancement or protrusion (non-conducting) Water tree increases local electric field stress 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 compared to water trees Cable System Failure Process 45 45

46 Diagnostics used in Challenging Areas Cable System Failure Process 46 46

47 Diagnostics in the Field Cable System Failure Process 47 47

48 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 48 48

49 Outline NEETRAC Overview CDFI Background/Overview Cable System Failure Process SAGE Concept Analytical Techniques & Failure Rates Diagnostic Testing Technologies Case Study: Roswell Diagnostic Accuracies Accuracies Really Matter The Things That Are Much Clearer Now CDFI Research Summary 49 49

50 SAGE Approach to Diagnostic Programs 50 50

51 Failures [#] Diagnostic Program Selection Action Generation Evaluation Increasing Failures Decreasing Failures SAGE Concept Time 51 51

52 Diagnostic Program Phases - SAGE Selection Data compilation and analysis needed to identify circuits that are at-risk for failure (at-risk population). Action Determine what actions can be taken on circuits based on the results of diagnostic testing. Generation Conduct diagnostic testing of the at-risk population. Evaluation Monitor at-risk population after testing to observe/improve performance of diagnostic program. SAGE Concept 52 52

53 Outline NEETRAC Overview CDFI Background/Overview Cable System Failure Process SAGE Concept Analytical Techniques & Failure Rates Diagnostic Testing Technologies Case Study: Roswell Diagnostic Accuracies Accuracies Really Matter The Things That Are Much Clearer Now CDFI Research Summary 53 53

54 Analytical Techniques & Failure Rates 54 54

55 Statistical Approach to Data Analysis Engineers are generally not fond of statistics We sometimes make decisions from test data using approaches that are overly simplified NEETRAC uses analytically rigorous techniques to enhance our approach to data analysis A quick tutorial. Analytical Techniques 55 55

56 Cable Moisture Content (Raw Data) 6 5 Moisture Content (%) Analytical Techniques 56 56

57 Histograms They show the distribution Mean StDev N Percent Normal Moisture Paper (%) Analytical Techniques 57 57

58 Boxplot No Assumptions About Distribution 6 5 Moisture Content (%) Median 50% of data lie above and below Mean 1 0 Analytical Techniques 58 58

59 Boxplot No Assumptions About Distribution Moisture Content (%) Upper Quartile 75% of data lie below Whisker Extends from the box to the Max or Min 1 0 Lower Quartile 25% of data lie below Analytical Techniques 59 59

60 Boxplot No Assumptions About Distribution 6 Moisture Content (%) Outlier A datum that cannot be considered to be part of the majority of the data 1 0 Analytical Techniques 60 60

61 Weibull Distribution Useful for Failure Data Probability, expressed as a %, of a sample having a lower value 95 Percent 20 5 Value being measured Moisture Content (%) 10.0 Analytical Techniques 61 61

62 Weibull Distribution Weibull - 95% CI Percent Shape Scale N % of samples are likely to have a moisture content below 2.2% Moisture (%) Analytical Techniques 62 62

63 Statistical Approach to Data Analysis Introduces rigor to the data analysis process Allows you to see true differences between data sets Allows you to combine data sets to gain further insight Reduces ambiguities Allows for extrapolation Recognizes that there are different types of data Allows for increased accuracy of the analysis when the data is sparse and imperfect Analytical Techniques 63 63

64 Composition of US MV system Installed Capacity (%) PILC HMWPE XLPE EPR TRXLPE UNKNOWN Analytical Techniques 64 64

65 Failure Rates 100 Peak at 140 Failure Rate [#/100 Miles/Year] Max: 140 Mean: 12 Upper Quartile: 8 Median: 3.5 Lower Quartile: Analytical Techniques 65 65

66 Failure Split Terminations 5.6% Unknown 1.1% Splices 37.1% Cable 56.2% Analytical Techniques 66 66

67 Each Utility is Different Utility C Cable Unknown Source A ccessories Percentage Analytical Techniques 67 67

68 Evolution of Failures (1) Log Cumulative Failures Type FAILURE TEST Failures Increasing Pre-Diagnostic Program Program Start Up Log Time (Days) Analytical Techniques 68 68

69 Evolution of Testing (1) No. of Tests (% of Year 3 Test Level) 400 Test Program Ramp Up Year 3 Level Year Analytical Techniques 69 69

70 Evolution of Failure Rates (1) 160 Test Program Ramp Up 3 Failure Rate (% of start) Inc Failure Rate Year 0 Gradient Year Analytical Techniques 70 70

71 Evolution of Testing (2) No. of Tests (% of Year 3 Test Level) 400 Test Program Ramp Up Year 3 Level Year Analytical Techniques 71 71

72 Evolution of Failures (2) Type FAILURE TEST Log Cumulative Failures Pre-Diagnostic Program Program Start Up Full Program Log Time (Days) Analytical Techniques 72 72

73 Evolution of Failure Rates (2) Test Program Ramp Up Full Program 140 Failure Rate (% of start) Year 0 Gradient Year Failures still rising just 4 5 not as fast 6 Analytical Techniques 73 73

74 QUESTIONS 74 74

75 Outline NEETRAC Overview CDFI Background/Overview Cable System Failure Process SAGE Concept Analytical Techniques & Failure Rates Diagnostic Testing Technologies Case Study: Roswell Diagnostic Accuracies Accuracies Really Matter The Things That Are Much Clearer Now CDFI Research Summary 75 75

76 Diagnostic Testing Technologies 76 76

77 Utility Use of Diagnostics Diagnostic Testing Technologies 77 77

78 Diagnostic Survey A survey of CDFI participants in 2006 was conducted to determine how diagnostics were employed. Survey was updated at the end of Survey results focused CDFI work on technologies currently used in the USA. Diagnostic Testing Technologies 78 78

79 Survey of Use of Diagnostics 27.8% 41.7% 30.6% No Testing Testing - one technique Testing - > one technique Diagnostic Testing Technologies 79 79

80 Survey of Use of Diagnostics More than one technique used No testing One technique used 4.0% Testing 25.0% No Testing 96.0% 75.0% No Testing Occasional use Regularly used Some testing Diagnostic Testing Technologies 80 80

81 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) Diagnostic Testing Technologies 81 81

82 Simple Dielectric Withstand 82 82

83 Simple Dielectric Withstand Test 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 83 83

84 Voltage EARLY Hold Entry Ramp Entry Withstand Test Process HOLD Voltages and Times for VLF covered in IEEE Std 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 t = 0 t Test Time Simple Dielectric Withstand 84 84

85 Examples of Withstand Units Simple Dielectric Withstand 85 85

86 VLF Waveforms RMS with 280feet XLPE Load Voltage (kv) Cosine-Rectangular SEBAKMT RMS with 280feet XLPE Load Sinusoidal Voltage (kv) Simple Dielectric Withstand 86 86

87 Test Sequences Withstand Test Outcomes Simple VLF Withstand to IEEE400.2 Levels Test time 30 mins Cumulative Length Tested in One Year (Miles) 140 Simple Dielectric Withstand 87 87

88 Dielectric Loss (Tan δ) 88 88

89 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 89 89

90 Voltage Dielectric Loss Test Process Loss measurement Time Tan δ 90 90

91 Tan δ Equipment Tan δ 91 91

92 Cable System Equivalent Cable system (cable, splices, and terminations) is reduced to simple circuit. Tan δ 92 92

93 Cable System Equivalent Tan δ 93 93

94 Tan δ Ramp Test Data Voltage [p.u.] Tan-delta [1e-3] Tip Up Mean Cycles Scatter (represented by Standard Deviation - IQR could be used) Time [min] Tan δ 94 94

95 Time Domain Reflectometry 95 95

96 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 96 96

97 TDR Principles Near End TDR Equipment L Joint Far End Joint TDR 97 97

98 TDR Equipment Accessed 9/30/ TDR 98 98

99 Lengths Tested PD Median 814 ft Tan D Median 485 ft Percent VLF Withstand Median 3500 ft Cable Length - log (ft) Panel variable: Technique Based on diagnostic data supplied to CDFI Measurements made with TDR TDR 99 99

100 Online Partial Discharge

101 Online Partial Discharge Test 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 (no customer outage required) Testing performed by a service provider. Assessment criteria are unique to each embodiment of the technology Measurements require sensor placement at multiple locations along cable circuit Online PD

102 Voltage Online PD Test Process Continuous measurement U 0 Time Online PD

103 Online PD Equipment Online PD

104 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 PD PD Online PD

105 Offline Partial Discharge

106 Test Description Offline Partial Discharge 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 analysis Offline PD

107 Voltage Offline PD Test Process Discharge measurement Time Offline PD

108 Offline PD Equipment Offline PD

109 PD Pulse 140 mv 180 pc Offline PD

110 PD Pulse Offline PD

111 PD Phase Resolved Pattern Offline PD

112 Offline PD (60 & 0.1Hz) Outcome Sequences A B C No PD PD Offline PD

113 Isothermal Relaxation Current Recovery Voltage

114 Test Description Isothermal Relaxation Current Measures the time constant of trapped charges within the insulation material as they are discharged. Discharge current is observed for minutes. Field Application Offline test that uses DC to charge the cable segment up to 1kV. Testing is performed by a service provider. IRC

115 Test Description Recovery Voltage 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

116 What does this mean for IRC & RV? Use limited to evaluation studies in the laboratory Possibly too sensitive for field use IRC & RV

117 Review 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) Diagnostic Testing Technologies

118 Issues to Keep in Mind About Diagnostics There is more to it than just the test Context Reporting Critical Levels

119 Clarifying Cable Diagnostics Diagnosis is defined as the art or act of identifying a disease from its signs and symptoms 1. A diagnosis would tell you what is wrong with your cable system (broken neutral, insulation voids, overheating connector etc.). Cable diagnostics today tell you whether your cable system is sick or not. Utilities typically ask diagnostics to tell them which parts of the cable system are sick. 1 Accessed 9/1/ Context, Reporting, & Critical Levels 119

120 Data Generation from Diagnostic Measurement Context is important Local Context Comparisons within one area Global Context Comparison with many tests Databases Standards 120 Context, Reporting, & Critical Levels 120

121 Diagnostic Measurements and Failures Symptoms are difficult to relate to future failures unless they are in the extremes. Good? Bad Probability No Failure Failure Diagnostic Measurement 121 Context, Reporting, & Critical Levels 121

122 Diagnostic Spectrum 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? 122 Context, Reporting, & Critical Levels 122

123 Data Driven How to Establish Levels Measurement data naturally segregate themselves into distinct classes Outcome Segregation Use service performance after measurement to segregate measurement data Darwinian Utilize all available knowledge and new data to update levels as needed 123 Context, Reporting, & Critical Levels 123

124 Prior Knowledge Darwinian Measure Monitor Initial Assessment Update Assessment Implement Prior knowledge used to generate initial assessments Time Levels updated based on service performance or other factors Accuracies cannot be determined until levels have stabilized 124 Context, Reporting, & Critical Levels 124

125 Outcome Segregation Measure Monitor Adjust Levels to Correlate To Failures Implement No prior knowledge of how measurements correlate with failures Levels determined after sufficiently long monitoring period Accuracies cannot be determined until levels have been set Time 125 Context, Reporting, & Critical Levels 125

126 Level-Based Reporting Systems Level-based (i.e. 1, 2, 3, Defer, Repair, Replace, Act, Don t Act etc.) reporting systems are increasingly common. Useful for condensing complex information into easier to understand categories. Categories typically represent the output from a black box analysis approach. Measurement System Data Experience Black Box Level 126 Context, Reporting, & Critical Levels 126

127 Caveats of Level-Based Systems 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 Further Study in terms of service performance? Comparisons / interpretation of different level-based reporting systems is difficult. Consistency of black box over time does the result mean the same thing next year? Cannot reassess later when you ve learned more. Stability = 0.5E-3, Tip Up = 20E-3 vs Further Study Required Need to associate meaning with the levels 127 Context, Reporting, & Critical Levels 127

128 Example Online PD Level % Percent % Provider Data Classes based on failures 3% Time to Failure (Years) Context, Reporting, & Critical Levels 128

129 Alternate Interpretation Original Level Alternate Class (based on probability of failure) << 3 < Class 18 has 6 times poorer endurance than Class 3 Class 89 is 5 times poorer than Class Context, Reporting, & Critical Levels 129

130 Data Driven Measure Monitor Analysis Accuracy Verified Implement Time Data distributions define levels multiple modes characterize different mechanisms Levels determined before monitoring phase Accuracies can be determined after monitoring 130 Context, Reporting, & Critical Levels 130

131 Example VLF Tan δ of Cable Systems Percent % of the measured data lie below a Tan Delta (at Uo) of 2e-3 Can segregated based on areas where the curves break Define areas that are normal and unusual Tan Delta at Uo (E-3) Context, Reporting, & Critical Levels 131

132 VLF Tip Up Data of Cable Systems Ins Class Filled PE Percent Percent TU TU Context, Reporting, & Critical Levels 132

133 - Cable System Treatment Unfilled Polyolefin Polyethylene Insulations Action Required Tan Delta (1e-3) Further Study No Action Tip Up (1e-3) Context, Reporting, & Critical Levels 133

134 Results of Tan δ Testing Service Failures [% of Tested] CDFI Data Classes based on data Action ACTION REQUIRED FURTHER STUDY NO ACTION FOT 1 10 Elasped Time between test and failure in service at May 09 (Month) 134 Context, Reporting, & Critical Levels 134

135 QUESTIONS

136 Outline NEETRAC Overview CDFI Background/Overview Cable System Failure Process SAGE Concept Analytical Techniques & Failure Rates Diagnostic Testing Technologies Case Study: Roswell Diagnostic Accuracies Accuracies Really Matter The Things That Are Much Clearer Now CDFI Research Summary

137 Case Study Roswell, GA November 2008 & January 2009 TDR Tan Delta Monitored Withstand Offline PD

138 Roswell Map Case Study: Roswell

139 SELECTION Case Study: Roswell

140 Roswell Background Info 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 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

141 Knowledge-Based System 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 Case Study: Roswell

142 Extruded Cable Diagnostics Case Study: Roswell

143 KBS Example Case Study: Roswell

144 Summary for Diagnostic Selection Replace Small Portion Replace Segment Replace Accessories 144 Diagnostic Technique TDR & Historical Records ONLY PD Offline PD Online Tan Delta Monitored Withstand HV DC Leakage VLF 60 Mins VLF 30 Mins VLF 15 Mins DC Withstand Action Scenario Have a shortlist of three techniques 144 Case Study: Roswell

145 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

146 Scenario Assessment before Testing Offline PD Typical Observations 0.5% fails on test, no customer interrupted 1 PD site / 1,000ft 40% of discharges in cable Qualitative Prediction (12 months) 0-1 fails on test 51 discharge sites 15 splices 1-2 failure within 12 months after test Historical Outcomes: 1-3 Failures Monitored Withstand Typical Observations < 4% (1000ft sections) fails on test, no customer interrupted 70% of loss tests indicate no further action Qualitative Prediction (12 months) 1-2 fails on test 3 assessed for further consideration 0-1 failure within 12 months after test Historical Outcomes: 1-3 Failures CDFI Research Section Offline PD CDFI Research Section Withstand Tan Delta Case Study: Roswell

147 ACTION Case Study: Roswell

148 Initial Corrective Action Options Replace splices only no detailed records assume 12 splices. Complete system replacement. Case Study: Roswell

149 GENERATION Case Study: Roswell

150 Overhead and Cabinet Terminations Case Study: Roswell

151 TDR Results Action_Required Segment 7 Phase 3 Segment 7 Phase 2 Segment 7 Phase 1 Segment 6 Phase 3 Segment 6 Phase 2 Segment 6 Phase 1 Segment 5 Phase 3 Segment 5 Phase 2 Segment 5 Phase 1 Segment 3 Phase 3 Segment 3 Phase 2 Segment 3 Phase 1 Segment 2 Phase 3 Segment 2 Phase 2 Segment 2 Phase 1 Segment 1 Phase 3 Segment 1 Phase 2 Segment 1 Phase 1 Further_Study TDR LEVEL No_Action Case Study: Roswell

152 Tan δ Monitored Withstand Case Study: Roswell

153 Monitored Withstand - Stability 18 Segments Tested Pass - Stable Loss Pass - Un Stable Loss 30 min test 60 min test Sequence of Lengths Tested (miles) 8 10 Case Study: Roswell

154 If this had been a Simple Withstand No Failures On Test 18 Segments Tested Length Tested (miles) 8 10 Case Study: Roswell

155 Importance Hierarchy of Tan Delta Tan δ Time Stability Tip Up [1.5U 0 0.5U 0 ] Tan δ [U0] Case Study: Roswell

156 Test Results - Local Perspective Action_Required Further_Study Case Study: Roswell LOSS LEVEL Segment 1 Phase 3 Segment 1 Phase 2 Segment 1 Phase 1 Segment 2 Phase 3 Segment 2 Phase 2 Segment 2 Phase 1 Segment 5 Phase 1 Segment 3 Phase 3 Segment 3 Phase 2 Segment 3 Phase 1 Segment 6 Phase 2 Segment 6 Phase 1 Segment 5 Phase 3 Segment 5 Phase 2 Segment 7 Phase 3 Segment 7 Phase 2 Segment 7 Phase 1 Segment 6 Phase 3 No_Action

157 Targeted Offline PD (VLF) Case Study: Roswell

158 Targeted Offline PD Test Segment 6 Phase A - 1 TDR B - 2 C - 3 Open symbols represent the anomalous TDR reflections A - 1 PD (Approx Positions) B - 2 C Distance from Cubicle 2 (ft) Case Study: Roswell

159 PD Inception local perspective VLF Test Voltage (kv) A - 1 B Position of PD (ft) PD in 1 of 9 splices PD in 1 of 7 splices C Position from Cubicle 2 (ft) Panel variable: Phase PD in 5 of 9 splices Case Study: Roswell

160 EVALUATION Case Study: Roswell

161 Evaluation after Testing Offline 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 9 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 10 months since test Case Study: Roswell

162 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

163 QUESTIONS

164 Outline NEETRAC Overview CDFI Background/Overview Cable System Failure Process SAGE Concept Analytical Techniques & Failure Rates Diagnostic Testing Technologies Case Study: Roswell Diagnostic Accuracies Accuracies Really Matter The Things That Are Much Clearer Now CDFI Research Summary

165 Diagnostic Accuracies What are diagnostic accuracies? Why do they matter?

166 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. 166 Diagnostic Accuracies 166

167 Where does the accuracy come from? Accuracies can be computed based on different approaches. Service Performance After Testing Comparison of Different Diagnostics In CDFI, the focus has been on comparing diagnostic data to service performance data. 167 Diagnostic Accuracies 167

168 Objective of Diagnostic Tests The 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 Target Population Good Components 168 Diagnostic Accuracies 168

169 Diagnostic Operation Applying the diagnostic will separate the population into: No Action Required group Action Required group No Action Required Action Required 169 Diagnostic Accuracies 169

170 Perspective Diagnostic technologies are designed to find anomalies in the field. Detecting the presence of an anomaly is not sufficient. The goal must be to detect an anomaly which leads to reduced reliability (failure in service) or compromised performance (severed neutrals stray voltage). In the CDFI, when a diagnostic indicates that a circuit is bad we interpret that to mean the circuit will fail in the near future. 170 Diagnostic Accuracies 170

171 Accuracies Variable time horizons of 1-8 years No Action Accuracy Action Accuracy Diagnostic Accuracy Dataset Diagnostic Accuracies

172 All Accuracies Diagnostic Accuracy No Action Accuracy Action Accuracy Overall Accuracy 172 Diagnostic Accuracies 172

173 Group Sizes Act vs. Not Act Tests [% of Total] Act Not Act

174 Outline NEETRAC Overview CDFI Background/Overview Cable System Failure Process SAGE Concept Analytical Techniques & Failure Rates Diagnostic Testing Technologies Case Study: Roswell Diagnostic Accuracies Accuracies Really Matter The Things That Are Much Clearer Now CDFI Research Summary

175 Why Do Accuracies Matter?

176 Diagnostic Program Benefit No Action Required Action Required Avoided Corrective Actions Avoided service failures Accuracies Really Matter

177 Diagnostic Program Loss No Action Required Action Required Future service failures Unneeded Corrective Actions Accuracies Really Matter

178 Cost [$] Total Diagnostic Program Cost Consequence Corrective Actions Total Diagnostic Program Cost Diagnostic Selection Accuracies Really Matter

179 Cost [$] Benefit and Loss BENEFIT Consequence Corrective Actions Diagnostic Selection Alternate Program 1 LOSS Alternate Program 2 Accuracies Really Matter

180 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

181 Uncertainty in Diagnostic Program Costs Cost [$] Consequence Program Cost Range Corrective Actions Selection Diagnostic Accuracies Really Matter

182 Uncertainty in Diagnostic Program Costs Cost [$] LOSS Program Cost Range BENEFIT Alternate Program 1 Accuracies Really Matter

183 Example Few Customers System Composition [% Bad] TRANSITION LOSS BENEFIT Prob_1 < > Diagnostic Accuracy 0.9 Accuracies Really Matter

184 Example Many Customers Population Composition [% Bad] LOSS TRANSITION BENEFIT Prob_1 < > Diagnostic Accuracy 0.9 Accuracies Really Matter

185 Summary Diagnostic programs include four cost elements (Selection, Diagnostic, Corrective Action, and Consequence) Benefit can be obtained from: Fewer corrective actions Improved reliability (fewer service failures) Modeling economics requires probabilistic approaches since many cost parameters are not known

186 QUESTIONS

187 Outline NEETRAC Overview CDFI Background/Overview Cable System Failure Process SAGE Concept Analytical Techniques & Failure Rates Diagnostic Testing Technologies Case Study: Roswell Diagnostic Accuracies Accuracies Really Matter The Things That Are Much Clearer Now CDFI Research Summary

188 The Things That Are Much Clearer Now CDFI Research

189 By Diagnostic Technique VLF DC Tan Delta PD On PD Off TDR IRC DAC Category No Use Occasional Standard Testing CDFI Research

190 CDFI Dielectric Withstand Dielectric Withstand

191 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. Need to look at both performance on test and service performance Explored the concept of Monitored Withstand tests. Dielectric Withstand

192 VLF Lab Program Dielectric Withstand

193 Overview Test program combining aging at U 0 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 Metric Survival during aging and testing Secondary Metrics Before and after each VLF application, PD at U 0 Between Phase A & B IRC, PD (AC 2.2U 0, DAC), Tan δ Dielectric Withstand

194 VLF Phases Samples Aging Voltage Aging Temperature VLF Voltage Type Status Phase A Service Aged XLPE U 0 Ambient Sine 0.1Hz Testing Complete Aging Complete Phase B Phase A Survivors 2U 0 45 C Cosine-Rectangular 0.1Hz Testing Complete Aging Underway Dielectric Withstand

195 Laboratory Setup Dielectric Withstand

196 VLF Units Cosine-Rectangular Sinusoidal Dielectric Withstand

197 Phase A U 0 & Ambient Temp Aging Sinusoidal VLF RMS with 280feet XLPE Load Voltage (kv) Dielectric Withstand

198 1: No Withstand 2: VLF 2.2U0 15 Min Withstand Testing Periods (variable durations) 3: VLF 3.6U0 120 Min 4: VLF 2.5U0 60 Min 5: VLF 2.2U0 120 Min Failures are the primary metric Aging Periods for evaluation Phase A End 6: 60 Hz 3.6U Min CDFI Regional Meeting Oct 13-14, 2009 Columbus, OH Dielectric Withstand

199 1: No Withstand 2: VLF 2.2U0 15 Min 3: VLF 3.6U0 120 Min 4: VLF 2.5U0 60 Min 5: VLF 2.2U0 120 Min 6: 60 Hz 3.6U Min T1 T2 T3 T4 Dielectric Withstand CDFI Regional Meeting Oct 13-14, 2009 Columbus, OH

200 1: No Withstand No Failures 2: VLF 2.2U0 15 Min No Failures 3: VLF 3.6U0 120 Min 4: VLF 2.5U0 60 Min 5: VLF 2.2U0 120 Min 3 VLF Failures No Aging Failures 2 VLF Failures No Aging Failures No Failures 6: 60 Hz 3.6U Min T1 T2 T3 T4 CDFI Regional Meeting Oct 13-14, 2009 Columbus, OH 2 60 Hz Failures No Aging Failures 200 Dielectric Withstand 200

201 Phase B 2U 0 & 45 C Ageing Cosine-Rectangular VLF SEBAKMT RMS with 280feet XLPE Load Voltage (kv) Dielectric Withstand

202 1: No Withstand 2: VLF 2.2U0 15 Min 3: VLF 3.6U0 120 Min 4: VLF 2.5U0 60 Min Phase B End 5: VLF 2.2U0 120 Min 6: 60 Hz 3.6U Min T1 T2 T3 T4 CDFI Regional Meeting Oct 13-14, 2009 Columbus, OH Dielectric Withstand

203 1: No Withstand No Failures 2: VLF 2.2U0 15 Min No Failures 3: VLF 3.6U0 120 Min 4: VLF 2.5U0 60 Min 5: VLF 2.2U0 120 Min 10 VLF Failures No Aging Failures 2 VLF Failures No Aging Failures No Failures Phase B End 6: 60 Hz 3.6U Min T1 T2 T3 T4 CDFI Regional Meeting Oct 13-14, 2009 Columbus, OH No 60 Hz EV Failures 2 Aging Failures Dielectric Withstand

204 Failures on Test When do they happen? > % < % Failures that would occur within IEEE Std test duration % % % Dielectric Withstand

205 Voltage Effect on Times to Failure 180 Phase I & II - Uo / RT ageing, Sine Phase III - 2Uo / 45C ageing, Cosine Time to 10% Failure (mins) Both curves show that higher voltage leads to increased failure rate on test More Failures Rated Voltage (Uo) Dielectric Withstand

206 VLF Test Program Summary Analysis of Phase A is complete. Phase B (2U 0 aging, 45 C Cosine-Rectangular) underway. Phases A & B show that no VLF exposed samples have failed under 60 Hz U 0 & 2U 0. Phase B tests shows two samples without VLF exposure failed during 60 Hz 2U 0. VLF failures on test: Less than 15 mins: 12 % (2 failures) mins: 71 % (12 failures) Dielectric Withstand

207 Withstand Field Experience Dielectric Withstand

208 Voltage EARLY Withstand Test Process HOLD Hold Entry Voltages and Times for VLF covered in IEEE Std Ramp Entry t = 0 t Test Time Dielectric Withstand

209 Withstand Testing Experience 100 Survivors [% of Total LengthsTested] IEEE Recommendation IEEE Range Time on Test [Minutes] Dielectric Withstand

210 Withstand Testing Experience 100 Survivors [% of Total Tested] Conductor Miles >2000 Conductor Miles IEEE VLF USA Utilities Time on Test [Minutes] Dielectric Withstand

211 Why the differences in Survivor Curves? Survivor curves differ between utilities because of: Test Voltage Tested Length Composition Early Failures on Test Dielectric Withstand

212 Length Distribution (Overall) 30 Median Length = 3500 ft Percent Wide variability in circuit lengths Circuit Length [Conductor ft] Dielectric Withstand

213 Expect Failures Increase with Time Failures on Test [% of Tested] Expectation Time on Test [Minutes] Dielectric Withstand

214 Early Phase Matters Early DC Ramp Entry Hold Failures on Test [% of Tested] % of failures on test occurred during Early phase Time on Test [Minutes] Dielectric Withstand

215 Early Phase - Cable & Joints Failures on Test [% of Initiation Failures] Type Cable Joint DC Ramp Entry U Test Voltage [kv] Regular Initiation Raw

216 Early and Hold Failure Mechanisms 20 Early VLF Hold Entry Hold Failures on Test [% of Total Tested] Multiple failure modes Time on Test [Min] Dielectric Withstand

217 DC or VLF Simple Withstand Length Adjusted Failures on Test [% of Tested Sections] % Difference 0.0 DC VLF Dielectric Withstand

218 Early and Hold Phases Length Adjusted Failures on Test [% of Tested Sections] DC 0.20 % 0.03 % due to VLF Difference between VLF and DC is primarily result of Early phase VLF FOT Type Hold Early Dielectric Withstand

219 Withstand Testing Experience 100 Survivors [% of Total Tested] What happens during Hold phase? 0 0 Many early failures Time on Test [Minutes] Dielectric Withstand

220 Test Performance for Different Utilities Failures on Test [% of 1000 ft. Segments] Utility A1 A2 D H I 1000 ft Length Adj Time on Test [Minutes] Dielectric Withstand