Challenges Associated with the Interpretation of Dielectric Loss Data ...

C.4.5 8th International Conference on Insulated Power Cables C.4.5

Jicable’11 – 19 – 23 June 2011, Versailles - France

Challenges Associated with the Interpretation of Dielectric Loss data from Power Cable System measurements J. Perkel (1), J.C. Hernández (2), R. N. Hampton (1), J. F. Drapeau (3), J. Densley (4), Y. Del Valle (1) 1 - NEETRAC, Atlanta, GA, USA, [email protected], [email protected],

yamille.delvalle @neetrac.gatech.edu 2 - Universidad de Los Andes, Mérida, Mérida, Venezuela, [email protected] 3 - IREQ, Varennes, Québec, Canada, [email protected] 4 - ARBORLEC, Mississauga, Ontario, Canada, [email protected] ABSTRACT Cable System Management requires an assessment of the health of the cables system. It is increasingly common for the assessment of aged cable systems to be made through the application of diagnostics measurements. A recent study has shown that VLF Tan Delta is perhaps the most commonly deployed cable system diagnostic. The practical use of this technique has been supported by the international standards IEEE400-2002 and IEEE400.2 (latest version). A key part of these standards is the guidance provided to a user that is detailed in the “Figures of Merit”. These enable users to make practical improvements to the cable system. To aid these decisions a series of criteria have been developed. The benefit of the criteria described here is that the process is rational, reproducible and transparent. The outcomes are supported by a probabilistic assessment of service performance.

KEYWORDS Diagnostic Techniques, Very Low Frequency (VLF), Tan Delta, Decision Tools

INTRODUCTION The use of dielectric loss measurements to estimate the “health” of cable system assets is becoming increasingly common: a study in 2009 conducted by the authors showed that dielectric loss tests were the most commonly implemented proactive diagnostic in North America [1]. The measurement of the dielectric loss on cable systems in the field has been discussed by many authors for frequencies ranging from 0.01 Hz (VLF – very low frequency) to 300Hz (DAC – Damped AC) [1 - 9]. However, these discourses have tended not to address the practical methods that might be used to determine the levels which define the “health classes” e.g. condition assessment results as: no action required, action recommended, and others. Therefore, this paper discusses the challenges associated with the interpretation of dielectric loss data from power cable system measurements at the different frequencies. The discussion is based on the following items or issues: • Limitations of current methods and standards (i.e.

IEEE Std. 400 Ed. 1). • Collation of more than 3000 individual dielectric loss

data from different cable systems in the field. • Selection and discussion of traditional and novel

diagnostic features such as absolute loss magnitude, changes in loss magnitude with voltage (Tip Up or Tip Down), temporal stability of loss magnitude for a specific voltage level, and stability of loss magnitude considering both time and voltage level.

• Interpretation of the local geographical context of loss data, e.g. effect of isolated high loss regions and neutral corrosion effects.

• Analysis of diagnostic features using Pareto principles establishes appropriate levels of performance for EPR, PILC & Polyethylene based cable system insulations.

• Assessment of the in service performance of the cable systems some years after the initial condition estimates.

VLF TAN DELTA MEASUREMENTS Tan δ measurements determine the degree of real power dissipation in a dielectric material (dielectric loss). A comparison relates this measurement to a known reference value for the type of dielectric measured. A judgment establishes the condition of the tested circuit based on how much the dielectric loss differs from the reference value. Reference values can be based on:

• Values measured on adjacent phases (A, B, C), • Values measured on cables of the same design

and vintage within the same location, • Values when new, • Industry standards, or an experience library.

Tan δ is most powerful if the specific cable and accessory components under test are known. This allows for a direct comparison between the measured value and: • The expected values for known materials /

components, • Previous measurements on the same circuit, or • Baseline values. Tan δ values are obtained by applying an AC voltage and measuring the phase difference between the voltage waveform and the resulting current waveform. Then this phase angle is used to resolve the total current (I) into its charging (IC) and loss (IR) components. The Tan δ is the ratio of the loss current to the charging current.

Figure 1: Equivalent Circuit for Tan δ Measurement and Phasor Diagram

Figure 1 shows an ideal equivalent circuit for a cable, consisting of a parallel connected capacitance (C) and a voltage dependent resistance (R). The Tan δ measured, at a frequency ƒ and voltage V, is the ratio of the resistive

AC

Excitation source0.1 Hz or 60 Hz

R CV

I

RI CI

VRI

ICI

δ

θRe

Img

Close and Return

mailto:[email protected]�








(IR) and the capacitive (IC) currents.

Table 1: Advantages and Concerns of VLF Tan δ

Advantages Concerns Energizing test equipment is small and easy to handle. Frequency dependency of Tan δ can be established. Tan δ is more sensitive at lower frequencies than at 60 Hz due to the reduced magnitude of the capacitive current. Can test very long cable circuits.

Testing voltage waveform may not be the same as the operating voltage. Frequencies lower than 0.01 Hz may cause space charge formation

REPORTING AND INTERPRETATION In principle, there are three types of dielectric loss data that may be reported:

• Tan δ normally reported as the mean of a number of sequential measurements (the median of these measurements may also be used).

• Differential Tan δ or Tip Up - normally reported as the simple algebraic difference between the means of a number of sequential assessments taken at two different voltages. The difference between medians may also be used.

• Tan δ stability - normally reported as a standard deviation of sequential measurements. The inter-quartile range (span of middle 50 % of the data) may also be used.

403020100

100

10

1403020100

PILC

Measurement Sequence

TD [

E-3]

XLPE

3.66.57.2

10.813.914.419.123.1

[kVrms]Voltage

Figure 2: Measured Tan δ data from Cable Systems in service

Figure 2 shows examples of measured Tan δ data from cable systems in service. The tests consisted of sequential Tan δ assessments at selected voltage levels. An extensive collation of VLF Dielectric Loss data has been assembled in a database that covers at least 22 discrete test areas and more than 3,700 data entries. In this collation the “filled” class refers to all cables with EPR (all types) or Vulkene insulation, “paper” refers to PILC cables, and “PE” refers to all cable with polyethylene

based insulations, including HMWPE, XLPE, and WTRXLPE insulations. These groupings have been adopted by recognising that in the vast majority of cases the user will not have the precise details of the cable system at the time of test – maps are often available but generally do not contain up to date information. These data are presented in a box and whisker format [6, 10] - Figure 3 includes three graphs: • “TD” – mean Tan δ measured at U0, • “TU 1.5-.5” – differential Tan δ measured at 1.5U0 and

0.5U0, and • “STAB” – stability measured by the standard deviation

for sequential measurements made at U0.

Figure 3: Dielectric Loss Feature Data segregated for Insulation Class

The authors strongly prefer limiting the maximum Tan δ test voltage to below that prescribed for withstand tests in IEEE 400.2™. Thus a Tip Up or Differential Tan δ is based on an upper voltage of 1.5Uo. This has two main advantages 1) the low risk of failure is further reduced and 2) the quality of the cable system is assessed before commitment to any withstand voltage.

The data in Figure 3 represent more than 3,300 segments with a mean length of 320 m (1,070 ft). The total length for this population exceeds 1120 km (700 conductor miles). The open circles represent the mean of the data sets. The horizontal lines within the boxes represent the median.

TRADITIONAL SUCCESS LEVELS In many instances, a condition assessment is attempted using Dielectric Loss Tan δ as the primary metric. This approach appears in the current data from IEEE Std. 400. The results of this approach are frequently problematic as the interpretation may be influenced by the length of the cable, the type of insulation and the presence / dispersion of high loss elements (terminations, highly water treed regions, or splices). Applying the criteria suggested in IEEE Std. 400 to the collated data available clearly demonstrates these issues. The standard suggests that cable systems with Tan δ > 4E-3 need to be replaced. However, inspection of the data shows that this implies that 47% of the systems measured require replacement. Not only does this appear to be an unreasonably high percentage of “bad” circuits, but the fact that most of these systems remain in service with a failure level of 11% indicates that the assessment accuracy may be improved. Furthermore many classified at the

9 8 7 6 5 4 3 2 1 0

P E P a p e r F i l l e d

2 0

0

- 2 0

P E P a p e r F i l l e d

1 0 0 8 0 6 0 4 0 2 0 1 0 5 0

STAB ( E - 3 )

I n s C l a s s

0

T U 1 . 5 - 0 . 5 ( E - 3 )

0

T D ( E - 3 )

0

Close and Return



intermediate levels fail in advance of those classed more severely. Thus, we conclude that the present IEEE Std. 400 criteria and those proposed in Annex F of IEEE 400.2 200.2 are too conservative at the most severe level and miss-classify at the mid level.

IEEE Std. 400 also notes that the critical levels will depend upon the insulation types used for the system. This contention is supported within Figure 3 for the basic insulation classes. The update of IEEE Std. 400.2 will address the critical levels, features, and insulation dependencies. It will revise much of the guidance presently in IEEE Std. 400. While many engineers focus on a Tan δ level, IEEE Std. 400 also suggests that multiple Tan δ features should be considered (i.e. Tan δ and Tip Up) in the assessment. Unfortunately, the standard does not provide guidance regarding how to make a decision. The use of multiple features, say Tan δ and Differential Tan δ, has proven useful in the analyses (Figure 3) conducted by the authors. In such a scheme, clarity and consistency in determining the critical levels is important.

ESTABLISHING CRITICAL LEVELS WITH MULTIPLE FEATURES In the past, engineers have tried to find “perfect” criteria that absolutely

a) One value lies more than three standard deviations from the mean or

separate the Tan δ values of components that go on to fail from those that do not. To do this requires a significant amount of service data on Tan δ and failures, which is difficult to acquire. Even then the multitude of ageing scenarios may preclude this. This is especially true for dielectric loss data that are typically collected by utilities. An alternative approach developed by the authors identifies critical dielectric feature levels that separate “usual” from “unusual” data. This is the classic Shewart or control chart approach, which uses the mean and standard deviation as a metric to define a “normal” value. In the simplest form, data are unusual if either:

b) Two sequential values are more than two standard deviations from the mean.

As a result, knowledge rules for Tan δ can now be further refined. The following sections describe the current database and its use in determining Tan δ critical diagnostic levels. This work relies on a hierarchy, established using step breakdown correlations [7], for Dielectric Loss features: • First Tier – STAB (measured by the Standard

Deviation) • Second Tier – Tip Up or Differential Tan δ • Third Tier –Mean Tan δ Figure 4 shows the distribution of Tan δ stability measurements at U0 for each insulation class (PE, Filled, and Paper). Stability, in this case, is assessed by the standard deviation of the data. As this figure shows, 80 % of the stability measurements are less than 0.35 E-3, 0.1 E-3, and 0.05 E-3 for Filled, Paper, and PE insulations, respectively. The choice of the 80th percentile relies on the Pareto Principle, which says that 80 % of the problems come from the worst 20 % of the population.

1.00.80.60.40.20.0

90

80

70

60

50

40

30

Stability (Std Dev) of Tan Delta @ Uo (E-3)

Pe

rce

nt

80

95

FilledPaperPE

Ins Class

Figure 4: Cumulative Distribution of all Cable System Stability Values at U0

100500-50

100

90

80

70

60

50

40

30

20

10

Tip Up of Tan Delta between 1.5 Uo & 0.5 Uo (E-3)

Pe

rce

nt

95

80

FilledPaperPE

Ins Class

Paper CablesRanges forBoundedDouble

Figure 5: Cumulative Distribution of all Cable System Tip Up Criteria – scale has been expanded to show

negative Tip Up on the Paper cables

10001001010.1

100

90

80

70

60

50

40

30

20

10

Mean of Tan Delta @ Uo (E-3)

Pe

rce

nt

95

80

FilledPaperPE

Ins Class

Figure 6: Cumulative distribution of all the Cable System Tan δ at U0

Figure 5 shows the distributions of Tip Up data for different ranges of Tip Up where Tip Up is the difference in Tan δ measured at 1.5U0 and 0.5U0. As many workers have noted, it is quite common for paper cables to show a negative tip up, however excessive negative values have been considered atypical in the same manner as unusually large ones. In this case the authors maintain the

Close and Return



same critical percentiles (80% & 95%) but employ them in a double bounded manner i.e. the 80% No Action class ranges from the 10th percentile to the 90th percentile.

Figure 6 shows the distributions of Tan δ measured at U0. In this case, the 80th percentile corresponds to Tan δ values of 40 E-3, 90 E-3, and 4 E-3 for Filled, Paper, and PE-based insulations, respectively. The power of this approach is that it utilizes the whole data set (cable systems that perform well and those that do not). Thus it is self updating as more data becomes available. However as the decision points are at the upper percentiles, the updating can cause apparently large changes in values. This is due to the shallow nature of the curves in the upper reaches, such that small changes in percentile result in large changes in value.

Table 2: Criteria for Condition Assessment of Cable Systems – collation of data to 2010

Condition Assessment

Tan δ Stability

at U0 [E-3]

Tip Up (1.5U0 – 0.5U0) [E-3]

Tan δ at

U0 [E-3]

PE, HMWPE, XLPE, & WTRXLPE No Action Required

<0.05 & <5 & <4

Further Study Advised

0.05 to 0.5 or

5 to 80 or

4 to 50

Action Required >0.5 >80 >50

EPR (conventional & un conventional) & Vulkene No Action Required

<0.35 & <10 & <40


0.35 to 1.5 or

10 to 180 or

40 to 125

Action Required

>1.5 >180 >125

PILC

No Action Required

<0.1 & -35 to 10 & <90


0.1 to 0.4

or

-35 to -50

or

10 to 100 or

90 to 200

Action Required

>0.4 <-50 or >100 >200

The approach used to determine the critical levels for diagnostic features from these data relies on the collated field data as of the end December 2010. Inspection of Figure 4 through Figure 6 show that in most of the cases (the exception being the mean Tan δ at U0 for Paper and Filled insulation) the data are not well modeled by simple distributions. In fact, the largest available data set (PE with >2,000 entries) indicates that there are suitable “breakpoints” between the distributions. By coincidence, these breakpoints are associated with probability levels of 80 % and 95 %, which are the same probability levels as found using other analysis techniques such as Pareto Analysis. Given this observation, these probabilities guide the Condition Assessments as shown below and they, together with the previously noted hierarchy, provide the basis for Table 2: • No Action Required encompasses the lowest 80 %

of the data

• Further Study encompasses the next lowest 15 % (80 % - 95 %) of the data

• Action Required encompasses the highest 5 % (95% -100%) of the data

The overall condition assessment of the circuit is most simply defined by the most “serious” condition of any of the dielectric loss features. In other words, if any one criterion indicates the circuit is “Action Required,” then the assessment is “Action Required” regardless of what the other two criteria indicate. Prioritizing or differentiating between circuits with the same overall assessment requires looking at all of the criteria and an approach to this is described in a later section. This scheme is very similar to the level-based systems used for other diagnostic techniques. However, in this case, the knowledge rules (i.e. the critical levels, the level criteria (80 % and 95 %), and the data base) are available to the user. The availability of these numeric data aids understanding by the user and permits re interpretation.

SERVICE PERFORMANCE Overall Many of the cable systems that have been measured have been followed up to determine the performance in the field of cable systems left in service. This elicits times to failure and the survival times for each of the action classes shown in Table 2. Figure 7 shows these times for PE based insulations (failure and survival) in Weibull format segregated by the action classes. The performance has been tracked to February 2011.

100.010.01.00.1

20

10

5

3

2

1

Elasped Time Feb 2011 (Month)

Failu

res o

f P

E B

ase

d I

nsu

lati

on

s (

%)

60

6

13

20

0.1 1

2

5

No Action RequiredFurther Study

Action Required

Overall Class

Figure 7: Diagnostic Performance Curves for Tan δ measurements on cable systems with PE based

insulations

Complimentary curves have been developed for Paper and Filled insulations. It is important to note that the classification is achieved using the “atypical” levels and not from the failures themselves. The failures and survivors are used to check the classification. The lower times correspond to the dielectric failures on test (FOT). The quality of the fit is also worth noting since the distribution fits the available data well, which leads to several significant conclusions.

These curves show that the likelihood of failure, if no

Close and Return



actions are performed after testing, follow the classifications from the “atypical” approach reasonably closely (i.e. a segment classed as “Action Required” has the highest probability of failure). Thus, these data show that there is a strong relationship between the cable system dielectric loss and subsequent service reliability. That is, an elevated Dielectric Loss feature (Tan δ, Tip Up or Unstable Tan δ) indicates a higher risk of failure in service

In common with almost all diagnostics, even the most severe classification is not necessarily an immediate “death sentence.” It clearly takes time for even the worst segments to fail. Thus we are identifying the propensity to failure not the failure itself. The vertical percentile lines in Figure 7 show the probabilities of failure for each condition assessment at selected times after test. Even after five years of service, only 20 % of the worst segments failed.

It is instructive to compare the performance achieved here with those from the present IEEE Std. 400 criteria and those proposed in Annex F of IEEE 400.2 2002. In this case the same dataset of polyethylene based cables used for Figure 7 was reinterpreted using these criteria and related to the true level of failures. The most striking feature from this exercise is the inversion of the service failure rates for the mid range and poorest performers: 19% & 12% respectively. In a well performing scheme it would be expected that the highest service failure rate would be associated with the poorest diagnostic class.

Case Study

1000100101

1000.0

100.0

10.0

1.0

0.11000100101

TD @ Uo

Tip

Up

1.5

Uo

- 1

.0 U

o

4 50 4 50

5

80

Area C Area W

Figure 8: Tan δ Results from 2007 Testing – open symbols indicate failures in service in 2010

Most texts on Tan δ testing note the benefits of comparison with previous measurements. In 2010 a unique opportunity arose to conduct measurements on a system that had been first investigated in 2007. Furthermore good service records were available for the intervening period. Two subdivisions were tested in 2007 using Tan δ Ramp and Tan δ Monitored Withstand. The cable system was a 15 kV unjacketed XLPE system installed in the mid-1970’s. The results of the original tests are shown in Figure 8. A full evaluation would require a representation with TD Stability, Tip Up and Mean TD. However for simplicity we content ourselves with a two dimensional representation here. Nevertheless, even without an all encompassing classification scheme, it is

clear that the performance of the sections in the upper right is different and poorer than the others. Therefore these would be the first candidates for any action.

During the 3+ years since the first tests were completed, three service failures occurred. The failures (circled areas in Figure 8 represent the areas that failed) were restricted to the sections that were identified as requiring attention in 2007. It is important to note here that these tests were conducted with a view to understanding the outcomes of diagnostic tests, thus the sections were not remediated but were left in service to establish the outcomes. If the criteria had been followed in 2007, then these service failures would have been avoided.

1000100101

1000.0

100.0

10.0

1.0

0.1

1000100101

2007

Mean TD @ Uo

Tip

Up

1.5

Uo

- 1

.0 U

o

4 502010

4 50

5

80

Figure 9: Comparison of 2007 and 2010 Tan δ Results

Figure 9 shows a comparison of the 2010 results with those obtained in 2007 for all segments tested both years. The changes in condition are summarised in Table 3. As can be observed the cable sections with the poorest conditions in 2007 showed the largest changes with the general trend to a poorer condition.

The circuits in Area C will continue in service and will be monitored whilst those in Area W will be replaced within the next few months.

Table 3 Summary of Service Failures and Changes in condition

No

Action

Required

Further Study

Action

Required

No per category - 2007 3 4 8

No failing in service 2007 to 2010 0 0 3

Change 2007 to 2010 on un failed circuits

Degraded but no change in

category 0 4 3

No Change in category 3 0 1

Improvement in category 0 0 1

Close and Return



MULTIVARIATE APPROACHES Most treatments of a Tan δ result uses a simple set of rules of the type set out in Table 2. Although this has been found to work well for the majority of cases (Figure 7) it is not so clear for the cases where two or more of the indicators lie in the upper range of the class when it might be argued that the diagnosis should be more severe than the simple levels would suggest. The authors recognised that the data set for polyethylene based cables was sufficiently large and had a high enough fidelity to enable this conjecture to be tested. Furthermore it was recognised that visualisation of the results from the testing would be assisted if it were possible to find a means by which the outcomes suggested by the disparate metrics could be combined to provide a single measure of health index.

The approach currently used for this work is Principal Component Analysis (PCA). This technique was chosen as this would provide a predictive model, guidance on the appropriate factors to combine, and would likely enable a physical meaning to be ascribed to the resulting composite factors – the Principal Components. The Principal Component approach identifies linear combinations of the factors that minimise the variance within the data. The advantage is that it now makes it possible to look for patterns in the data when there are more factors than can be handled by simple graphical means.

Figure 10: Empirical Distribution for the Magnitude of Principal Components (>1500 cases plotted) for PE

based Cable Systems

In addition to the classic metrics (Table 2) the authors took the opportunity of examining combinations and ratios which some workers had suggested would give improved effectiveness. Extensive modeling showed that the combinations and ratios did yield acceptable decisions. However the classical metrics, or small modifications thereof, provided much improved effectiveness whilst still providing physical meaning for the decisions. The outcome of this initial evaluation is shown in Figure 10. This approach achieves a good description of the data by encompassing almost 97% of the variance

Figure 10 shows the combined magnitude of the three Principal Components for all the cases of PE cable systems considered here. The percentage or rank position is given by the Y axis values and in practice might conveniently be regarded as a Health Index [11]. The symbols represent some selected data from Figure 9. The solid square symbol is a poor performer in 2007 that failed after approximately 27 months of further service. This would, in 2007, be described as being within the poorest 4% of all systems upon which we have data.

The open round and diamond symbols are measurements on single phase segments in 2007 & 2010 respectively. The magnitude of their Principal Components has been calculated and their positions have been plotted relative to all the available data. In this case the cable segment has degraded from a poorest 15% ranking in 2007 to a poorest 9% ranking in 2010. This is a 2 to 3% rate of degradation in rank position or health index per year in service. Inspection of mid range and lower ranks shows that, as might be expected, there is a degradation in rank here as well, but the rate is much lower: approx 0.33 to 0.5% loss of position per year. Information of this type is invaluable to an asset manager when determining the most appropriate route forward.

DECISION TOOLS The progress evidenced in Figure 10, and Table 2 shows that collated service test data enables a series of decision tools to be constructed that aid implementation in the field. The authors have successfully used a number of such tools, which have ranged from printed analysis sheets and data forms to spreadsheet Principal Component Tools. However to make the tools useful it is necessary to provide guidance on what the detailed activities might be for the cases (Further Study – 15% and Action Required – 5%) that require the user to respond. The current activity guidance is detailed below: Actions following a Further Study diagnosis might include • Review data for a rogue measurement in the

sequence – most common in the first acquisition • Check insulation type so that correct assessment

table is used • Re-clean terminations & repeat measurements • compare with previous tests or other results from

other phases of this cable • Conduct IEEE400.2 Standard (30 mins) VLF

Withstand whilst monitoring Tan δ • Place on “watch list” Actions following an Action Required diagnosis might include • Activities described for Further Study • Conduct IEEE400.2 Standard (60 mins) VLF

Withstand whilst monitoring Tan δ • Retest in near future • Place on “watch list” & schedule remedial actions for

the circuit

43210

100

80

60

40

20

0

Magnitude of the Principal Components (Arb Units)

Pe

rce

nt

Occu

re

nce

54321

100

95

90

85

80

75

70

Magnitude of the Pr incipal Components (A rb Units)

Pe

rce

nt

Occu

re

nce

80

95

performanceIncreasingly poor

Close and Return



ACKNOWLEDGEMENTS The authors gratefully acknowledge the useful discussions with many of the engineers involved within the Cable Diagnostic Focused Initiative and the financial support of a large number of utilities in North America and the US Department of Energy under award number DE-FC02-04CH11237. They are especially endebted to the utilities who took the time to contribute data from the field.

CONCLUSIONS This paper has shown that considerable progress has been made in the practical implementation of Tan δ diagnostic tests for medium voltage (MV) cable systems.

The authors have shown how a significant amount of data may be collated to garner data driven assessment criteria for cable systems. The use of a single set of percentiles enables a consistent and relatable set of criteria (eg the action limit in PILC cable systems can be understood in the same manner as that for PE based cable systems) to be established for all insulation types.

The analyses have been formatted so that they may be readily used in the field to provide real-time guidance on the appropriate decisions that a user might take to proactively manage their cable system asset.

Despite the progress that has been made, there are some challenges that remain to be clearly identified:

The multivariate approach should be extended to Filled and PILC cable systems. The PILC system presents a challenge due to the common finding of a negative tip up and the rarity of large negative values.

Current embodiments of field measurement systems provide real time representation of voltage effects and simple diagnostic factors (mean & standard deviation). Given the progress in the analysis presented here, it is clear that the user would benefit from: • The calculation of more diagnostic features, for both

Tan δ and Capacitance, such as Tip Ups and Tip Up Ratios, etc.

• A graphical representation of the Tan δ & Capacitance evolution with time.

REFERENCES 1. Diagnostic Testing of Underground Cable Systems

(Cable Diagnostic Focused Initiative), DOE Award No. DE-FC02-04CH11237, December 2010

2. L.A. Dissado, and J.C. Fothergill, “Electrical degradation and breakdown in polymers,” IEE Materials and Devices series 9, Peter Peregrinus Ltd., London, 1992.

3. J. Densley, “Aging Mechanisms and Diagnostics for Power Cables - An Overview,” IEEE Electrical Ins Mag, vol 17, no 1, pp 14-22, Jan/Feb 2001.

4. IEEE Standard 400.2-2004™, IEEE Guide for Field Testing of Shielded Power Cable Systems Using Very Low Frequency (VLF), 2005.

5. P. Werelius et al “Dielectric Spectroscopy for Diagnosis of Water Tree Deterioration in XLPE Cables,” IEEE Trans. on Dielectrics and Electrical Ins, vol 8, no 1, pp 27-42, Mar 2001.

6. R.N. Hampton et al; “Practical Issues Regarding The Use Of Dielectric Measurements To Diagnose The Service Health Of MV Cables”; JICABLE07, Versailles France, June 2007

7. J. C. Hernandez-Mejia et al; “Characterization of Ageing for MV Power Cables Using Low Frequency Tan-delta Diagnostic Measurements,” IEEE Trans on Dielectrics and Electrical Ins, Vol. 16, Issue 3, pp 862-870, June 2009

8. J. F. Drapeau et al, “Time Domain Spectroscopy (TDS) As A Diagnostic Tool For MV XLPE Underground Lines,” JICABLE07, Versailles, France, June 2007

9. IEEE Standard 400-2001™, IEEE Guide for Field Testing and Evaluation of the Insulation of Shielded Power Cable Systems,. 2002.

10. G. E. P. Box, W. G. Hunter, and J. S. Hunter, Statistics for Experimenters, John Wiley & Sons, Inc., New York, NY, USA, 1978.

11. E Dorison et al,” Health Index”; JICABLE07, Versailles, France, June 2007

Close and Return

Challenges Associated with the Interpretation of Dielectric Loss Data ...

Documents

Transcript of Challenges Associated with the Interpretation of Dielectric Loss Data ...