CIRED 22nd International Conference on Electricity Distribution Stockholm, 10-13 June 2013

Paper 0466

CORRELATION OF ACCELERATED AGEING PHENOMENA AND LONG-TERM CABLE

PERFORMANCE

Paul Brigandi Simon Sutton Stephen Cree

The Dow Chemical Company – USA Dow Chemical Company Ltd – UK Dow Europe GmbH
[email protected] [email protected] [email protected]

are a good indication of ultimate cable performance and

ABSTRACT expected lifetime. The qualification of MV cable is
Utilities, cable makers and industry experts generally possibly unique inasmuch as following an extended wet
agree that the results of medium voltage cable ageing protocol, such as CENELEC or AWTT testing,
qualification testing are a good indication of ultimate the cables are electrically broken down and their retained
cable performance and expected lifetime. Here we focus electrical strength measured. Although such qualification
on the development of a new enhanced cable insulation tests differentiate poor performing materials from those
material for distribution cables. While qualification tests with superior performance, this paper will show that
can differentiate poor performing materials from those correlating qualification data to life expectancy is much
with superior performance, it will be illustrated that more complex. Data from laboratory development to
correlating such qualification data to life expectancy is commercialization of the enhanced water tree retardant
much more complex. The data presented show that the insulation C4202 are used to illustrate the complexity of
enhancements observed during material development in correlating different test procedures with expected cable
water treeing behavior and breakdown strength life. Throughout the study, the previous generation
measured in the short-term 30 day laboratory tests led to insulation material HFDB-4202 (B4202) is used as a
a moderate 15% improvement in qualification results. benchmark.
However, significantly larger increases in the time-to-
failure and characteristic life time of 400% for cables in EXPERIMENTAL
the ACLT test were recorded. It would be expected that
similar enhancements would be achieved for cables Water Tree Growth Rate
under service conditions in the field. Water tree growth rate testing was conducted according
to ASTM D6097-01a for “Relative Resistance to Vented
INTRODUCTION Water-Tree Growth in Solid Dielectric Insulating
Materials.” Samples of the insulation compound were
In the 1970s, unjacketed high molecular weight moulded and crosslinked into 2” diameter disks with a
thermoplastic polyethylene and cross-linked polyethylene moulded-in defect produced by a conical needle with a
(XLPE) cables began failing prematurely with water precise tip radius (3 μm). The samples were then aged in
treeing being associated with the cable failures [1-3]. U-tubes filled with a 0.01 molar NaCl solution. The
While successive improvements to insulation, semicons samples were aged for 30 days under a constant AC
and jacketing have enhanced the performance of XLPE voltage of 5 kV at 1 kHz.
cable, the development of water tree retarding XLPE
insulation in the early 1980s is arguably the most Accelerated Cable Aging Tests
significant advancement [4]. Acceleration of cable aging can be achieved by imparting
higher electrical stress, high temperature, and the
Since its introduction in 1983, Dow Electrical & presence of a wet environment as accelerating aging
Telecommunications’ patented water tree retardant factors. Several aging times are generally required and
crosslinked polyethylene insulation (TR-XLPE) has provide a means to monitor the deterioration of electrical
replaced conventional XLPE to become the predominant breakdown strength as a function of aging time. In the
medium voltage (MV) insulation for primary distribution Americas and in parts of Asia, a 1-year AWTT according
cables in the U.S. and Canada. This insulation is also to ICEA S-694-94 is utilized as a part of the qualification
widely used in many countries, notably Germany and process and in Europe, the 2-year CENELEC HD 605
Russia in Europe, as well as other developed and aging protocol has been adopted more broadly. Two of
developing regions. the accelerated tests presented here are the AEIC
Accelerated Water Tree Test (AWTT) [5] and the
In 2010, Dow Electrical & Telecommunications launched Accelerated Cable Life Test (ACLT) [6].
the latest generation of TR-XLPE insulation, DOW
ENDURANCE™ HFDC-4202 (C4202) and the first full Accelerated Water Tree Test
scale qualifications tests have now been completed. In AWTT, twelve samples of 1/0 15 kV cable core (no
Utilities, cable makers and industry experts generally jacket) with 4.4 mm (175 mil) insulation wall thickness
agree that the results of MV cable qualification testing are aged in PVC conduits. The specimens are subjected

CIRED2013 Session 1 Paper No 0466

CIRED 22nd International Conference on Electricity Distribution Stockholm, 10-13 June 2013

Paper 0466

to a 60 Hz test voltage. A 14 day preconditioning was Utilizing the water tree growth rate test, water trees were
conducted to ensure that the test is not influenced by the grown and measured from a point defect moulded into
presence of volatile byproducts from the crosslinking the insulation material test sample. As depicted in
reaction. The first three samples are tested for AC Figure 1, water trees formed during this test are
breakdown (ACBD) after the initial preconditioning. The significantly smaller with the new enhanced insulation,
other nine samples are then tested for a year under ‘wet’ approximately 50% shorter than the incumbent insulation
conditions. Three samples each are removed after 120, grade. As shown in Figure 2, the C4202 TR-XLPE
180, and 360 days and tested for ACBD. The grows smaller and more constrained water trees
requirements are 26, 22.8, and 15 kV/mm respectively for compared to conventional XLPE and B4202 TR-XLPE.
three aging times.
Short-term laboratory scale tests such as water tree
Accelerated Cable Life Test growth rate and AC breakdown are convenient tools for
The ACLT evaluates comparative life of combinations of quickly screening multiple material formulations for
insulating and semiconductive shielding material designs performance. Such tests have been shown to correlate
of 15 kV cable specimens in water-filled tanks. with cable performance [7].
Completion of the ACLT is defined as failure of all of the
test specimens. The test was performed on 1/0 15kV (a) (b) (c)
cable core with 4.4 mm (175 mil) insulation wall
thickness. A continuously applied test voltage of 4Vg (4
times rated voltage to ground of a cable rated for a 15 kV
phase-to-phase voltage) and current loading to achieve a
conductor temperature of 90 °C in air was utilized until
all test specimens failed. Time-to-failure statistical
analysis is commonly achieved through fitting the Figure 2. Cross-sectional micrographs of water tree
population of failure data to a Weibull (or log-normal) growth after 30-day wet aging according to ASTM
failure distribution function. The Weibull plot compares D6097-01a Water Tree Test of (a) XLPE, (b) previous
the cumulative failure distribution versus time-to-failure generation TR-XLPE (B4202), and (c) new generation
on a special coordinate system which will yield a straight TR-XLPE (C4202)
line which is characterized by two parameters: “Eta” (the Full scale cable testing, according to the AWTT
characteristic time, where 63.2% of the population has qualification protocol, has been completed by cable
failed) and “Beta” (the slope or shape factor). manufacturers in North America for the new insulation
material. Testing according to the European CENELEC
RESULTS AND DISCUSSION HD 605 protocol is still ongoing due to the longer (two
Research during the development of C4202 involved year) qualification period.
studying the influence of the material composition on its 60
XLPE
AC Breakdown Strength, kV/mm

resistance to growing water trees and electrical 50 B4202

breakdown strength over different timescales. Data were C4202
benchmarked against conventional XLPE and the existing 40
generation water tree retardant product B4202.
30
Water Tree Comparison
ASTM D6097; 30 day Aging
20
0.6

10
0.5
Water Tree Length (mm)

0
0.4
Cyclic Aged 120 Day 180 Day 360 Day
0.3 AWTT Aging
Figure 3. AC Breakdown strength following AWTT
0.2 ageing protocol.
0.1
The average high-voltage AC breakdown strength results
0 are shown in Figure 3 and compared to typical values for
XLPE TR XLPE C4202
B4202 (the previous generation material) and XLPE
Figure 1. Water tree length comparison after 30-day insulation. The enhanced TR-XLPE material offers
wet aging according to ASTM D6097-01a Water Tree retained wet-aged electrical strength at the end of testing
Test of XLPE, previous generation TR-XLPE (B4202) of over 34 kV/mm (800 V/mil) average: this easily
and new generation TR-XLPE (C4202) surpasses the AWTT industry requirement of 15 kV/mm
(380 V/mil), at the end of one year. The breakdown

CIRED2013 Session 1 Paper No 0466

CIRED 22nd International Conference on Electricity Distribution Stockholm, 10-13 June 2013

Paper 0466

strength of C4202 has increased by 15% over B4202 after

360 days of wet aging.

This increase in retained breakdown strength may seem

small compared to the 50% decrease in water tree length
observed in Figure 1; however, there are significant
differences between the tests. Most notably the AWTT
protocol uses tap water which is separated from the
insulation by the semicon layers, whereas ASTM D6097-
01a uses a saline solution to accelerate tree growth [8]
from artificial defects intimately in contact with the water
source. It is generally considered that under service
conditions XLPE cables which do not employ a radial
water barrier (so called “wet design”) will likely last 25
years, whereas those using water tree retardant insulation
will surviving in excess of 40 years [9,10]. This is
Figure 4. 4,4 ACLT Weibull Failure Distribution
corroborated by very few reported failures due to
Analysis comparing the previous and new TR-XLPE
insulation degradation in cables made with TR-XLPE in
formulations throughout development.
close to 30 years of service operation [9]. The higher
retained breakdown strength following AWTT testing of
Table 1 summarises the results of the laboratory scale
cables made with water tree retardant insulation
tests and full cable tests comparing the change in
compared to conventional XLPE is generally seen as an
performance for C4202 versus B4202. From the
indication of the superior life expectancy (40 versus 25
perspective of a cable owner/operator the results in Table
years) for this insulation. As such a 15% improvement in
1 are highly positive. The laboratory test results translate
retained breakdown strength following AWTT testing for
to improved retained electrical breakdown strength
the new enhanced TR-XLPE may not be considered
following wet ageing of full scale cables. Though it may
striking and its impact on service life might be predicted
be considered that the 15% improvement in retained
to be limited.
breakdown strength following AWTT qualification
testing is moderate, it has been shown that this small
The results in Figure 4 show time-to-failure data in the
change leads to a large increase in the time-to-failure and
ACLT test for cables manufactured with different
characteristic life time; a worst case increase of four
insulation materials but the same grade of semicon. Even
times the results for the incumbent insulation has been
during the development phase for the new enhanced
achieved in the ACLT test.
insulation (Prototype and Pilot C4202), the characteristic
time-to-failure is more than double the existing grade
Table 1. Change in various performance measures
(B4202); approximately 750 days versus 350 days. The
between enhanced water tree retardant insulation
Prototype and Pilot plant C4202 made insulations have
(C4202) and previous generation (B4202)
the same chemical composition as the commercial grade,
but have been manufactured in small volumes for Test Laboratory Scale Tests Full Cable Test
research purposes and lack the high level of cleanliness
Retained
of the commercial grade. It is therefore surprising that Water tree Retained
electrical Time to
these materials still outperform the commercial B4202 length after electrical
breakdown Failure
Result water tree breakdown
insulation by over 100%. growth rate strength in
strength in Characteristic
cable Life
test plaques
testing
To date cables made from the commercially available
Change
C4202 insulation have shown no failures after 805 days, for C4202 Decreased Increased Increased Increased
as such a Weibayes analysis [11] has been used where it compared 50% 25% 15% 400%
is assumed that a failure occurred at the current time and to B4202
the shape parameter is assumed similar to that of the
prototype and pilot samples; and will thus be a “worst These data suggest that the ACLT protocol gives a more
case” estimate (dashed line) of a 1-parameter Weibull meaningful indication of cable life than one or two year
distribution. Applying this method an estimated qualification tests (AWTT or CENELEC). These
characteristic time-to-failure close to 1500 days is standards do nevertheless differentiate high performing
achieved. This is a four-fold increase in life compared to cables from those with lower performance. It is also
the previous insulation grade. important to note that the performance of a cable is
determined by the combination of insulation, semicons
and quality of manufacture.

CIRED2013 Session 1 Paper No 0466

CIRED 22nd International Conference on Electricity Distribution Stockholm, 10-13 June 2013

Paper 0466

During AWTT testing bow tie trees tend to initiate in the performance, correlating these results to life expectancy
first 120 days of ageing and subsequently not increase is much more complex and the ACLT protocol is a useful
significantly in size over the rest of the ageing period. tool to aid utilities in this regard. It would be expected
Figure 3 could be interpreted as mirroring this behavior, that similar enhancements would be achieved for cables
the initial fall in breakdown strength accompanies the under service conditions in the field.
growth of bow tie trees, which then tend to self-limit, and
then only decreases slowly with further ageing. There is REFERENCES
little difference in the size of number of bow tie trees in
B4202 and C4202; consequently there is little difference [1] Lawson, J.H. and Vahlstrom Jr., W. “Investigation
in the breakdown strength of the materials following the of Insulation Deterioration in 15 kV Polyethylene
AWTT ageing protocol. XLPE on the other hand grows Cables removed from Service, Part II.” IEEE Trans.
larger and more highly branched bow tie trees over the PAS Vol. 92, March/April, 1973, pp. 824-831.
same time frame, which in turn, leads to a higher [2] Bahder, G., Katz, C., Lawson, J.H., and Vahlstrom
localized stress and larger impact on the breakdown Jr., W. “Electrical and Electromechanical Treeing
strength compared to TR-XLPE materials. Effects in Polyethylene and Crosslinked
Polyethylene Cables.” IEEE Trans. PAS Vol. 93,
In contrast the ACLT protocol records the time to failure May/June 1974, pp. 977-986.
and is consequently a measure of the resistance to initiate [3] Eichorn, R.M. “Engineering Dielectrics”, Vol. II A,
vented trees and the rate at which these trees grow. The pp. 355 - 444, 1983.
cable fails once these vented trees either bridge the [4] P.J. Caronia, A. Mendelsohn, L.H. Gross, J.B.
insulation or grow beyond some critical length and Kjellqvist, 2006, “Global Trends and Motivation
initiate electrical trees that progress through the Toward the Adoption of TR-XLPE Cable”, AVO
insulation. In either case the cable fails when the NZ Conference.
degraded insulation can no longer hold the applied [5] Association of Edison Illuminating Companies,
voltage resulting in a phase-to-earth fault. C4202 retards 1994, “Specifications for A Thermoplastic and
the growth of vented trees more effectively than B4202, Crosslinked Polyethylene Insulated Shield Power
as seen in Figure 1 and Figure 2, and this is reflected in Cables Rated 5 through 46 kV”, 10th Edition, New
the much longer characteristic time to failure (Figure 4). York, (AEIC CS5-94)
The ACLT mirrors the real-world ageing mechanism of [6] R. Lyle and J.W. Kirkland, 1981, “An Accelerated
cables in service, and is thus arguably a better indication Life Test for Evaluating Power Cable Insulation”,
of the manner in which cables will perform in service IEEE Trans. PAS, Vol. PAS, Vol 100, no. 8, pp.
than the AWTT test. 3764-3771.
[7] J.O. Bostrom, J.O. Marsden, N. Hampton, U.H.
The AWTT (or European equivalent CENELEC) and Nilsson, 2003, “Assessment of cable performance
ACLT measure different aspects of cable ageing. as measured by a variety of accelerated ageing
Utilities generally base specifications on the AWTT or tests”, Proc. Inter. Conf. Insul. Power Cables,
CENELEC standards to ensure a desired level of Versailles, pp 572-578.
performance (and life expectancy). However as has been [8] J.P. Crine, J. Jow, 2000, “Influence of frequency on
shown above, the ACLT method is a better discriminator water treeing in polyethylene”, Electrical Insulation
of anticipated life. Longer cable life leads to lower life and Dielectric Phenomena, 2000 Annual Report
cycle costs [12] for the asset owner. Conference, vol.1, no., pp.351-354.
[9] B. Richardson, G. Stano, S. Ramachandran, 2008,
“Assuring Distribution Cable System Reliability”,
CONCLUSIONS
CIGRE Canada Conf. on Power Systems.
The development of an improved MV cable insulation, [10] EPRI, “Distribution Cable Research Digest 2000”,
from laboratory to commercial deployment, has been EPRI Publication BR-110693 (Available to
reviewed. It has been shown that: members only).
• Improvements in data seen in short-term laboratory [11] W. Wang; D.R. Langake, 2001, “Comparing two
tests translate to increased performance in full size designs when the new design has few or no failures-
cables. is the new design better than previous one?”,
• These enhancements in water treeing behavior and Reliability and Maintainability Symposium, 2001.
breakdown strength lead to moderate improvement in Proceedings, pp 322-325
the qualification results. [12] S. Sutton, “A Life Cycle Analysis Study of
• However a minimum four-fold increase in the time-to- Competing MV Cable Materials”, Proc. 21st Inter.
failure and characteristic life time for cables in the Conf. Elec. Distrib, 6-9 June 2011, Frankfurt.
ACLT test is observed.
While qualification tests can differentiate poor
performing materials from those with superior

CIRED2013 Session 1 Paper No 0466