GTRI/DOE Disclaimer. Cable Diagnostic Focused Initiative. Outline. CDFI Background. ICC Education Session. Nigel Hampton Rick Hartlein Joshua Perkel

Transcription

1 GTRI/DOE Disclaimer ICC Education Session Cable Diagnostic Focused Initiative Nigel Hampton Rick Hartlein Joshua Perkel Spring 9 Meeting 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-FC-4CH7 Outline CDFI Background/Overview Cable System Failure Process SAGE Concept Case Study: Roswell Diagnostic Accuracies Diagnostic Testing Technologies Accuracies Really Matter The Things We Know Now That We Did Not Know Before Selecting a Diagnostic Testing Technology Summary CDFI Background Rick Hartlein 4

2 Why do we need diagnostics? Composition of US MV system 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. Cable Failures per Year Installed Capacity (%) Always remember that we are talking about the cable SYSTEM, not just cable. PILC HMWPE XLPE EPR TRXLPE UNKNOWN CDFI Background/Overview CDFI Background/Overview 6 Failure Split Overview Unknown.% Terminations.6% In the CDFI, NEETRAC worked with 7 utilities, manufacturers and diagnostic providers to achieve the objective of clarifying the concerns and defining the benefits of diagnostic testing. Splices 7.% Cable 6.% Phase has almost exclusively focused on aged medium voltage systems. This is the largest coherent study of cable system diagnostics anywhere. CDFI Background/Overview 7 CDFI Background/Overview 8

3 NEETRAC Members Non NEETRAC Members Diagnostic Providers CDFI Supporters Dept of Energy Participants American Electric Power HV Technologies Ameren Hydro Quebec Cablewise / Utilx CenterPoint Energy Con Edison Cooper Power Systems IMCORP NRECA PacifiCorp (added mid ) Pacific Gas & Electric (added Jan 6) Duke Power Company PEPCO Exelon (Commonwealth Edison & PECO) Oncor (TXU) First Energy Prysmian Florida Power & Light Public Service Electric & Gas Georgia Tech Tyco / Raychem GRESCO HDW Electronics Southern California Edison Southern Company HV Diagnostics Southwire CDFI Background/Overview 9 CDFI Background/Overview CDFI - Primary Activities CDFI Activities ) Technology Review ) Analysis of Existing (Historical) Data ) Collection and Analysis of Field (New) Data 4) Verification of VLF Test Levels ) Defect Characterization 6) Develop Knowledge Based System 7) Quantify Economic Benefits 8) Reports, Update Meetings and Tech Transfer Seminars Analyses are data / results driven Analysis Lab Studies CDFI Field Studies Dissemination CDFI Background/Overview CDFI Background/Overview

4 CDFI Activities CDFI Activities Analysis Lab Studies CDFI Field Studies Dissemination Lab Studies Value / Benefit VLF Withstand Georgia Power Handbook Accuracies Tan δ Duke Publications VLF Withstand Tan δ PD Utility Data PD Meetings IEEE Std Work Knowledge Based Systems Industry CDFI CDFI Background/Overview Test Time Test Voltage Forensics Time Stability Voltage Stability Non-Uniform Degradation Neutral Corrosion Calibration Phase Pattern Feature Extraction Classification CDFI Background/Overview 4 Georgia Power XLPE Jkt & UnJkt Conductor Miles CDFI Activities Field Studies Duke XLPE & Paper Jkt & UnJkt 9 Conductor Miles CDFI Activities Analysis 89, Conductor Miles Offline PD (.Hz) Offline PD (6Hz) Tan δ Monitored Withstand Offline PD (.Hz) Tan δ Monitored Withstand Value / Benefit Accuracies Utility Data IEEE Std Work Knowledge Based Systems Evans Macon Roswell Charlotte * Cincinnati Clemson Morresville CDFI Background/Overview Economic Model SAGE DC Withstand Offline PD Online PD Tan δ VLF Withstand 4 Omnibus 4. VLF Survey Expert System Application CDFI Background/Overview 6 4

5 CDFI Activities Utility Data CDFI Activities Utility Data Con Ed Com Ed PPL Alabama Power Keyspan FPL PEPCO PG&E ONCOR Ameren DC Withstand Online PD VLF Withstand Offline PD (6Hz) Online PD Tan Delta VLF Withstand Offline PD (.Hz) Tan Delta Online PD Offline PD (.Hz) Tan Delta Offline PD (6Hz) VLF Withstand Offline PD (6Hz) Offline PD (.Hz) Online PD VLF Withstand Offline PD (6Hz) Online PD Tan δ Offline PD (6Hz) Online PD Offline PD (6Hz) CDFI Background/Overview 7 CDFI Background/Overview 8 Data Type Diagnostic Technique DC Withstand Monitored Withstand PD Offline PD Online Tan δ VLF Withstand Dataset Sizes Laboratory [Conductor miles] Field [Conductor miles] 78, ,8 Benefits from Diagnostic Programs Decreasing failures associated with diagnostics and actions Log Cumulative Failures Program Initiated IRC. - Service Performance ALL 89, 8 9 Time [Days] CDFI Background/Overview 9 CDFI Background/Overview

6 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. Where we are today (). Diagnostics work they tell you many useful things, but not everything.. Diagnostics do not work in all situations.. 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.. 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 CDFI Background/Overview Where we are today () Overview 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..important to have correct expectations diagnostics are useful but not perfect! In the CDFI, NEETRAC worked with 7 utilities, manufacturers and diagnostic providers to achieve the objective of clarifying the concerns and defining the benefits of diagnostic testing. CDFI Background/Overview 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. CDFI Background/Overview 4 6

7 Failures by Equipment How things fail and what fails have a big impact on the selection of diagnostics Cable System Failure Process Rick Hartlein Disbursement of Failures (%) Cable - all (%) Splice - all (%) Terminations - all (%) Unknown - all (%) Cable System Failure Process 6 Major Cable Components Defect Types in Extruded Cables Conductor Conductor or Strand Shield Insulation Insulation Shield Metallic Shield/Neutral Jacket (Recommended) Cable System Failure Process 7. Cavity at shield(s). Cavities due to shrinkage. Insulation shield defect 4. Contaminant (poor adhesion). Protrusions at shield(s) 6,7 Splinter/Fiber 8. Contaminants in insulation or shields Cable System Failure Process 8 7

8 Conversion of Water to Electrical Trees Defect Types in Extruded Cable Accessories Electrical tree growing from water tree Acts as a stress enhancement or protrusion (non-conducting) Water tree increases local electric field Water tree also creates local mechanical stresses If electrical and mechanical stresses high enough electrical tree initiates Electrical tree completes the failure path rapid growth Cable System Failure Process 9 Cable System Failure Process Diagnostics used in Challenging Areas 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 8

9 Diagnostic Program Phases - SAGE SAGE Approach to Diagnostic Programs Nigel Hampton 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 4 Failures [#] SAGE at Work Selection Action When to deploy diagnostics Increasing Failures Decreasing Failures Generation Evaluation Cable System Performance Commissioning Operational Stress Condition Assessment Time (Years) 4 SAGE Concept Time 6 9

10 Context is important Data Generation from Diagnostic Measurement Local Context Comparisons within one area Case Study Roswell, GA November 8 & January 9 Nigel Hampton Global Context Comparison with many tests Databases Standards TDR Tan Delta Monitored Withstand Offline PD SAGE Concept 7 8 Roswell Map SELECTION Case Study: Roswell 9 Case Study: Roswell 4

11 Roswell Background Info. 98 vintage XLPE feeder cable, kcmil, 6 mils wall, jacketed. Knowledge Based Selection System Failures have occurred over the years no data on source Recently experienced very high failure rates of splices on this section: 8 failures / 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 4 Selecting a Diagnostic Technology 4 Summary for Diagnostic Selection Economic Details prior to testing Action Scenario DC Withstand VLF Mins VLF Mins Diagnostic Technique VLF 6 Mins HV DC Leakage Monitored Withstand Tan Delta PD Online PD Offline TDR & Historical Records ONLY Complete System Replacement $,, approx Complete Splice Replacement $6, Test time (determined by switching) - 4 Days Selection Costs $, Splice Replacement 7 Days Retest after remediation Day Replace Small Portion Replace Segment Replace Accessories Monitored Withstand, Offline PD and VLF ( mins) offer economic benefit over doing nothing. Have a shortlist of three techniques Case Study: Roswell 4 Case Study: Roswell 44

12 Scenario Assessment before Testing Offline PD If,ft is tested.% fails on test, no customer interrupted site /,ft (median) 4% discharges in cable Estimate fails on test discharge sites cable, accessories splices < failure in months from test Monitored Withstand If,ft is tested <4% fails on test, no customer interrupted 7% of loss tests indicate no further action Estimate < fails on test assessed for further consideration by loss. failure in months from test ACTION 4 Case Study: Roswell 46 Initial Corrective Action Options Replace splices only no detailed records assume splices. Complete system replacement. GENERATION Case Study: Roswell 47 Case Study: Roswell 48

13 Overhead and Cabinet Terminations Monitored Withstand Case Study: Roswell 49 Case Study: Roswell If this had been a Simple Withstand Monitored Withstand - Stability 8 Segments Tested Pass - Stable Loss Pass - Un Stable Loss No Failures On Test min test 8 Segments Tested 6 min test 4 6 Length Tested (miles) Sequence of Lengths Tested (miles) 8 Case Study: Roswell Case Study: Roswell

14 Test Results - Local Perspective Test Results Global Perspective Tip Up in Tan Delta {.Uo -.Uo} (e-) Numbers indicate Segment ID's Length Along Feeder (ft) Measurement Stability STABLE UNSTABLE 6 7 Tan Uo (e-)... Stability STABLE UNSTABLE Splice Error Range of instability in monitored withstand Termination Damage 6 Panel variable: Phase. Tip Up.Uo -. Uo (e-) Case Study: Roswell Case Study: Roswell 4 Targeted Offline PD Test Segment 6 Phase A - Targeted Offline PD TDR B - C - Open symbols represent the anomalous TDR reflections A - PD (Approx Positions) B - C - Distance from Cubicle (ft) 4 4 Case Study: Roswell Case Study: Roswell 6 4

15 PD Inception local perspective VLF Test Voltage (kv) A - B - PD in of 9 splices C PD in of 9 splices Probability of Splice Inception (%) Position of PD (ft) PD in of 7 splices Position from Cubicle (ft) Panel variable: Phase PD Inception (kv) EVALUATION Case Study: Roswell 7 Case Study: Roswell 8 Evaluation after Testing Offline PD,ft actually tested Estimate discharge sites 6 cable, 9 accessories 6 splices < failure in months from test Actual 7 discharge sites cable, 7 accessories splices failure 4 months from test Monitored Withstand,ft actually tested Estimate fails on test assessed for further consideration by loss. failure in months from test Actual fails on test 6 assessed for further consideration by stability, tip up & loss failure (cable) months from test 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. 9 Case Study: Roswell 6

16 Performance of Diagnostics Performance evaluation primarily focuses on diagnostic accuracy. Diagnostic Accuracies Nigel Hampton 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. 6 6 Diagnostic Accuracies 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 Diagnostic Operation Applying the diagnostic will separate the population into: No Action Required group Action Required group But the diagnostic is imperfect... 6 Diagnostic Accuracies 64 Diagnostic Accuracies 6

17 Perspective Bad Means Failure Accuracies 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). Diagnostic Accuracy No Action Accuracy Action Accuracy In accuracy estimates we have used failures in service and interpreted the diagnostics as Bad Means Failure Dataset 6 Diagnostic Accuracies 66 Diagnostic Accuracies All Accuracies Diagnostic Accuracy [%] Diagnostic Testing Technologies Nigel Hampton Action Accuracy No Action Accuracy Overall Accuracy Diagnostic Accuracies

18 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. Utility Use of Diagnostics Different voltage sources may be used to perform the same measurement. DC 6 Hz. AC Very Low Frequency (VLF) AC Damped AC (DAC) Diagnostic Testing Technologies 69 Diagnostic Testing Technologies 7 Diagnostic Survey Survey of Use of Diagnostics A survey of CDFI participants in 6 was conducted to determine how diagnostics were employed. Survey was updated at the end of % 4.7% Survey results focused CDFI work on technologies currently used in the USA..6% No Testing Testing - one technique Testing - > one technique Diagnostic Testing Technologies 7 7 8

19 Survey of Use of Diagnostics Lengths Tested More than one technique used No testing One technique used. 7. PD Median 84 ft Tan D Median 48 ft. 4.% Testing.% Percent... VLF Withstand 7. Median ft Panel variable: Technique No Testing No Testing Occasional use Regularly used Some testing Diagnostic Testing Technologies 96.% 7.% 7... Cable Length - log (ft) Based on diagnostic data supplied to CDFI Diagnostic Testing Technologies 74 Context Diagnostic Context Data Generation from Diagnostic Measurement OK Not Proven either way NOT OK Local Context Comparisons within one area Global Context Comparison with many tests Databases Standards Extreme conditions are easy to decide what to do about. What to do about the ones in the middle? How to define the boundaries? SAGE Concept 7 Diagnostic Testing Technologies 76 9

20 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 6 Hz. AC VLF AC Damped AC Testing may be performed by a service provider or utility crew. Voltage INITIAL Withstand Test Process t = HOLD Voltages and Times for VLF covered in IEEE 4. t Test 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 Time Diagnostic Testing Technologies 77 Diagnostic Testing Technologies 78 VLF Test Voltages Test Voltage (kv) 6 4 Cosine-rectangular Sinusoidal Data Generation from Diagnostic Measurement Variable Peak Voltage (kv) Peak Voltage (kv) Peak Voltage (kv) RMS Voltage (kv) Use Acceptance Installation Maintenance Acceptance RMS Voltage (kv) Installation RMS Voltage (kv) Maintenance Cable Rating (kv) Diagnostic Testing Technologies 79 Diagnostic Testing Technologies 8

21 Test Sequences Withstand Test Outcomes 6 7 Simple VLF Withstand to IEEE4. Levels Time of failure in mins for failures > mins Local Context Comparisons within one area Cumulative Length Tested in One Year (Miles) 4 Diagnostic Testing Technologies 8 Diagnostic Testing Technologies 8 Separation with Simple VLF Outcomes Failures on Test [% of Tested] Area is clearly different from the others. Overall Area 4 Global Context Comparison with many tests Databases Standards 8 Diagnostic Testing Technologies 84

22 Withstand Testing Experience Test Performance for Different Utilities Survivors [% of Total LengthsTested] Time on Test [Minutes] 97 Conductor Miles > Conductor Miles. Conductor Miles IEEE Recommendation IEEE 4. Range 6 7 Failures on Test [% of ft Segments].. Utility A A D I ft Length Adj.... Time on Test [Minutes] 6.7%.% 4.4%.%. Diagnostic Testing Technologies 8 Diagnostic Testing Technologies 86 Service Failures [% of Total Tested]. Service Experience Test Conditions 6 Conductor Miles Time to Failure % [Days] U 47 U Days 67 Days 47 Days Time to Failure [Days since test] Time to Failure % [Days] Days % % 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: 6 Hz. AC VLF AC Damped AC Testing may be performed by a service provider or utility crew. Diagnostic Testing Technologies 87 Diagnostic Testing Technologies 88

23 Dielectric losses - Tan δ: I V Dielectric Loss (Tan δ) The cable insulation system is represented by an equivalent circuit In its simplest form it consists of two parameters; a resistor and a capacitor [IEEE Std. 4] When voltage is applied to the cable, the total current will be the contributions of the capacitor current and the resistor current I R I I R C I tan( δ ) = DF = = I IC ωrc C δ Cable System Equivalent Cable system (cable, splices, and terminations) is reduced to simple circuit. Diagnostic Testing Technologies I R θ V 89 Cable System Failure Process 9 Tan δ Test Data Data Generation from Diagnostic Measurement Tan-delta [e-] Voltage [p.u.]....7 Tip Up Mean Time [min] Scatter (represented by Standard Deviation - IQR could be used) 4 Diagnostic Testing Technologies 9 Diagnostic Testing Technologies 9

24 Tan δ Data for EPR Cable Systems Highest Concern Local Context Comparisons within one area Tan Delta (e-) Lowest Concern Voltage (kv) 9 Diagnostic Testing Technologies 9 Diagnostic Testing Technologies 94 Segments within a Feeder Lengths within a Locality Tip Up in Tan Delta {.Uo -.Uo} (e-) Stable STABLE UNSTABLE Phase = 6 7 Tan Delta (e-) Length Along Feeder (ft) 4 6 Length (ft)

25 Testing at Reduced Voltages. Regression 9% C I 9% PI.. 4 Global Context Comparison with many tests Databases Standards Uo [E-].... CI: Confidence Interval PI: Prediction Interval.. Uo [E-]...7. Diagnostic Testing Technologies 97 Cool Wall 98 Tan Delta - 6 Tan δ Interpretation Based on 8 Conductor Miles Action Required Further Study No Action 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. - - Tip Up - Diagnostic Testing Technologies 99 Diagnostic Testing Technologies

26 TDR Principles Online Partial Discharge Near End TDR Equipment L Joint Far End 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. Joint Field Application Online test that does not require external voltage supply. Testing typically only be performed by a service provider. Assessment criteria are unique to each embodiment of the technology Diagnostic Testing Technologies Discharge Occurrence Data Generation from Diagnostic Measurement No PD PD Diagnostic Testing Technologies 4 Diagnostic Testing Technologies 6

27 Distribution of PD along Lengths ft. portion of sample feeder Mixture of different PD levels for different sections and accessories. Local Context Comparisons within one area Cable Section Accessory No PD PD Diagnostic Testing Technologies Diagnostic Testing Technologies 6 Diagnostic Results (Overall) Accessory 6 Conductor Miles Cable 4.8% 4.9%.%.% 4.4% 4.8%.%.4% Global Context Comparison with many tests Databases Standards 66.6% 68.% Diagnostic Testing Technologies 7 Diagnostic Testing Technologies 8 7

28 Offline Partial Discharge Test 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: 6 Hz. AC service provider VLF AC utility crew Damped AC utility crew Diagnostic Testing Technologies 9 Amplitude [mv] Near End Detection Equipment P P P PD PD Activity Time P-P Time P-P P P P Time [μs] Far End Diagnostic Testing Technologies PD Pulse Data Generation from Diagnostic Measurement 4 mv 8 pc Diagnostic Testing Technologies Diagnostic Testing Technologies 8

29 PD Phase Resolved Pattern PD Magnitude.E- 4.E- 6 -.E- -6 Amplitude [pc] PD Level (pc) Max Limit for 97's production Individual Measurement from the field -.E- 8 Phese [Deg] - 6. Max allowed for current production.7.. PD Measurement Voltage (Uo) 4 PD Charge Magnitude Distributions XLPE Local Context Comparisons within one area Percent 4 Apparent Charge Magnitude [pc] 6 Diagnostic Testing Technologies Diagnostic Testing Technologies 6 9

30 PD Inception Voltage 8 6 XLPE 4 Percent Apparent Inception Voltage [U].4 Global Context Comparison with many tests Databases Standards 7 Diagnostic Testing Technologies 8 Location of PD Offline PD Test Sequence 6.6% of PD sites detected in accessories Conductor Miles Testing sequence for 6, ft. T ermination 6.% Cable 9.4% Splice 4.% No PD PD Diagnostic Testing Technologies 9

31 PD Location PD Sites per Length 8 6 Terminations Median =.96 PD Sites/ ft Approx. PD Site/ ft Percent Cable & Splices PD Sites per feet Location [% of Circuit Length] 7 9 Diagnostic Testing Technologies Isothermal Relaxation Current Recovery Voltage Test Description 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 kv. Testing is performed by a service provider. Test Description Similar to IRC only voltage is monitored instead of current Field Application Offline test that requires initial charging by DC source up to kv. Testing is performed by a service provider. Diagnostic Testing Technologies 4 Diagnostic Testing Technologies

32 Survey of Use of Diagnostics Combined Diagnostics 7.8% 4.7% Multiple degradation mechanisms mean that two diagnostics are often better than one.6% No Testing Testing - one technique Testing - > one technique Diagnostic Testing Technologies 6 Multiple Diagnostics.% Category Tan Delta / PD VLF / Tan Delta Data Generation from Diagnostic Measurement 7.% Diagnostic Testing Technologies 7 8 Diagnostic Testing Technologies

33 Monitored Withstand - Data Monitored Withstand Data - Elbow Tan-delta [e-].8 Optional Ramp Up gives Tan-delta.7 vs. Voltage Stability of Tan-delta monitored through the minute withstand Time [min] Voltage [p.u.]...7 Tan-delta [e-] Voltage [p.u.]....7 Elbow Failure Hampton Leas Segment HL Tan-delta vs Voltage Before Failure 4 Time [min] After Failure Failure Tan-delta Stability 6 7 Diagnostic Testing Technologies 9 Cool Wall Monitored Withstand Global Context Comparison with many tests Databases Standards Withstand Test Outcomes UNSTABLE High TU Poor Stability High Loss Monitored VLF Withstand to IEEE4. Levels Time of failure in mins Cumulative Length Tested in One Year (Miles) 7 Simple VLF Withstand to IEEE4. Levels 9 Diagnostic Testing Technologies Teste

34 Cost [$] Diagnostic Program Costs Accuracies Revisited Why do they matter? Josh Perkel Corrective Actions Consequence Total Diagnostic Program Cost Diagnostic Selection Accuracies Really Matter 4 Recall the Example... Incorrect Diagnosis No Action Required Action Required No Action Required Action Required Avoided Corrective Actions Avoided service failures Future service failures Unneeded Corrective Actions Accuracies Really Matter 6 Accuracies Really Matter 4

35 Cost [$] Benefit and Loss Considerations Diagnostic program economic calculations are based on ability to predict future failures. BENEFIT Consequence Corrective Actions Diagnostic Alternate Program LOSS Alternate Program Total diagnostic program cost is more sensitive to certain elements than others. Failure Rate Diagnostic Accuracy Failure Consequence Selection Accuracies Really Matter 7 Accuracies Really Matter 8 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 Accuracy Over Time Bad Means Failure Accuracy [%] No Action Required Accuracy Action Required Accuracy? System Changes Additional Aging Increased Load Time [Years] Accuracies Really Matter 9 4 Accuracies Really Matter

36 Probabilistic Approach Tan δ Probabilistic Approach - PD Failures in Service [% of Tested Segments]. Action_ Action_ Act Act No Action No Action 7.4%.9% Percent PD Class No PD PD FOT Elasped Time at Oct 8 (Month). Days Between Test & Failure Accuracies Really Matter 4 Accuracies Really Matter 4 By Diagnostic Technique VLF DC Tan Delta The Things We Know Now That We Did Not Know Before PD On PD Off TDR IRC DAC Category No Use Occasional Standard Testing 4 44 Diagnostic Testing Technologies 6

37 Dielectric Withstand CDFI Dielectric Withstand Josh Perkel Withstand techniques are most widely used diagnostic in the USA. Most utilities use VLF (either sine or cosine-rectangular) in their withstand programs. Test duration and voltage are critical to performance on test and in service. Explored the concept of Monitored Withstand tests. CDFI Research 4 CDFI Research 46 Length Distribution (Overall) Median Length = ft Length Effects Comparison of withstand failure on test rates must include length adjustments. ft. Percent Failure Wide variability in circuit lengths ft. ft. ft. ft Circuit Length [Conductor ft] 7 84 Censored CDFI Research 47 CDFI Research 48 7

38 Utility I Hybrid System Effect of Test Voltage Failures on Test [%] Length Adjustment Feet Feet NONE Performance at longer test times can be predicted. Length Weighted Average FOT Mins.7% 6 Mins.%... Time on Test [Minutes]. 7.% 4.%.4% Survivors [% of Tested] IEEE Rec. Level PILC Extruded Mixed (PILC and Extruded) NEETRAC Extruded... Test Voltage (U = Rated Voltage) CDFI Research 49 CDFI Research. VLF Lab Program Josh Perkel Overview Test program combining aging at U 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 Between Phase A & B IRC, PD (AC.U, DAC), Tan δ CDFI Research CDFI Research 8

39 : No Withstand : VLF.U Min Withstand Testing Periods (variable durations) : No Withstand : VLF.U Min No aging failures for any condition No Fails 4 Survive No Fails 4 Survive : VLF.6U Min 4: VLF.U 6 Min Aging Periods Phase A End : VLF.6U Min 4: VLF.U 6 Min VLF Fails Survive VLF Fails Survive : VLF.U Min 6: 6 Hz..6U. Min Failures are the primary metric for evaluation : VLF.U Min 6: 6 Hz..6U. Min No Fails 4 Survive 6 Hz. Fails Survive CDFI Research T T T T4CDFI Research 4 Time of Failure on Test Voltage of Failure on Test Percent of ft Samples Failing 99 Scale Failure Censor 8 Correlation Shape.468 No Failures Before mins After 6 mins 6 Mins Max Time 6 Percent 99 Shape Scale F ailure Censor 8 Correlation More failures occur at higher test voltages Time on VFL Test (min) Testing Voltage (Uo) CDFI Research CDFI Research 6 9

40 Failure Analyses - Trees & Defects in Cables Failed Samples D C B A DEFECT LARGE WATER TREE MEDIUM WATER TREE SMALL WATER TREE VLF Test Program Summary Analysis of Phase A is complete. Phase B (U aging, 4 C Cosine Rectangular) underway. Phases A & B show that no VLF exposed samples have failed under 6 Hz Uo & Uo. Phase B tests showed two samples without VLF exposure failed during 6 Hz Uo. All failures occurred at the appropriate time. i.e. within the VLF testing periods. Distance Along Cable (ft) 8% (4 out of ) of VLF failures between and 6 mins CDFI Research 7 CDFI Research 8 KBS Selecting the right diagnostic is not easy. Selecting a Diagnostic Technology Knowledge-Based System Nigel Hampton 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 9 6 Selecting a Diagnostic Technology 4

41 Knowledge Based Systems Extruded Cable Diagnostics 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 6 Selecting a Diagnostic Technology 6 KBS Example Short Listing of Diagnostic Approaches Experts Recommending a Diagnostic Technique (%) RECOMMENDED BY MOST EXPERTS RECOMMENDED BY FEWEST EXPERTS Discharge Monitored Withstand Simple Withstand Dielectric Discharge Historical Data Simple Withstand Monitored Withstand Simple Withstand Selecting a Diagnostic Technology 6 64 Selecting a Diagnostic Technology 4

42 Impact of Remedial Action Hybrid Cable System Most service failures occur in Accessories Usual remediation is by replacement of cable sections System Component Portion [%] Service Failure Rate Age [yrs] PE Medium - EPR 4 Low - Paper High 4 - Selecting a Diagnostic Technology 6 Hybrid Cable System Most Recommended Selecting a Diagnostic Technology 4 Expert Recommendation (%) Dielectric Discharge History & TDR Monitored W ithstand Simple Withstand Replace Replace Accessories Accessories Dielectric Discharge History & TDR Monitored W ithstand Simple Withstand Replace Replace Segment Segment Dielectric Discharge History History & & TDR TDR Monitored Monitored Withstand Withstand Simple Simple Withstand Withstand Replace Replace Small Small Section Section Summary Rick Hartlein 67 What we have learned about diagnostics (). A developing database of field failure diagnostic data shows that different diagnostic techniques can provide some indication about cable system condition.. Even if the diagnostics themselves are imprecise, diagnostic programs can be beneficial.. 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.. Utilities HAVE to act on ALL replacement/repair recommendations to get improved reliability. 68 Summary

43 What we have learned about diagnostics () 6. PD, VLF, DC and Tan δ & VLF withstand tests detect problems in the field and can be used to improve system reliability. 7. It is very difficult to predict whether or not the problems/defects detected by PD and Tan δ will lead to failure in the short/medium term. What we have learned about diagnostics ().Interpretation of PD measurements is more complex than interpretation of Tan δ measurements..irc & RV are particularly difficult to deploy in the field. 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..tan δ & PD measurements require interpretation to establish how to act. Summary 69 Summary 7 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. 7 Summary 4