140 IEEE Trensactions on Electrical Insulation Vol. 26 No. 1 , February l9Ql

Influence of Insulation Morphology, Impurities and Oxidation on some

Electric Properties of Cables

Jean-Pierre Crine, Serge P6lissou and Jean-Luc Parpal

Institut de recherche d’Hydro-Qukbec, Varennes, Canada

ABSTRACT It is shown that the ac breakdown strength and dielectric loss of actual unaged and aged transmission and distribution ca- bles vary with the insulation crystallinity, contamination and oxidation. The influence of these parameters is particularly im- portant near the conductor shield and it varies with aging. The influence of curing on cable properties is also briefly discussed.

INTRODUCTION

T is well known that electrical properties, such as break- I d own strength, dielectric losses and conductivity of di- electrics for cable applications depend to great extent on morphology [l, 21, oxidation [3] and contamination [l, 41. Recent studies have shown that the electric breakdown strength of polyethylene (PE) and crosslinked polyeth- ylene (XLPE) samples increases with their crystallinity (or density). However, there are few da ta obtained with actual cables aged in service. I t is also known that oxida- tion increases the tan6 value of P E and XLPE but little is known on the influence of aging (including oxidation and impurity diffusion) on the electrical properties of actual cables in service.

The objective of this paper is to present some prelim- inary measurements performed on unaged and field-aged transmission and distribution cables that show the influ- ence of insulation morphology, contamination and oxida- tion on their ac breakdown strength and dielectric losses.

I t is also shown that some of these parameters often have contradictory effects on the electrical properties of cables. In other words, aging may have few beneficial and many detrimental consequences, depending on the aging conditions and on the type of cable used.

SAMPLE PREPARATION

HE cable samples studied include a wide variety of T manufacturing and curing processes, as well as very different aging conditions [l, 51. The distribution cables have either HMW-PE or steam-cured XLPE insulation and they were aged in service from 6 to 11 yr. The trans- mission cables were mostly XLPE cables with steam and dry-cured insulation; in that case we had pairs of unaged and aged (either laboratory or field-aged) cables. In order to perform the radial analyses described later we peeled off ribbons ( x 100 to 200 p m thick) from the cables by tool-cutting on a lathe (see [l] for details). Great atten- tion was paid to cleanliness during sample preparation. The impurity contents and distributions were determined on these ribbons by neutron activation analysis [l, 51.

CRYSTALLINITY, IMPURITY AND OXIDATION MEASUREMENTS

RYSTALLINITY and density were estimated a t vari- C ous locations along the ribbons by DSC and density- gradient column measurements, respectively. More de- tails can be found in [1,5]. As shown in [l] the density distributions of cable insulation is typical of the curing

0018-9367/91/0200-140$1.00 @ 1991 IEEE

IEEE Transactions on Electrical Insulation Vol. 26 No. I , February 1991

... O p 0 o Q D o

Insulation thidtness (mm)

Figure 1. Typical radial distribution of insulation density in unaged and laboratory aged dry-cured XLPE cables.

E 0

2 U

0 5 15 Insulation thickness (mm)

Figure 2. Typical radial distribution of calcium (measured by neutron activation) in a laboratory aged dry- cured XLPE cable insulation.

process (if any), of cable aging and size. For example, Figure 1 shows the typical density distributions of un- aged and laboratory aged dry-cured XLPE transmission cables. I t appears that aging increases the density (i.e. crystallinity). Near the conductor shield of transmission cables, a density gradient is observed and the gradient is more important with dry-cured than with steam-cured XLPE [l]. Interestingly, in many cases, the density gra- dient does not correspond to a crystallinity increase [5]; this is thus not only a morphological modification of the

l-

. I

l-

-----

L

141

11

0 2 4 6 8 10 12 14 16 18 20 Insulation thicknw (mm)

Figure 3. Radial distribution of density of laboratory aged cable (same as in Figure 2).

r 2

U Q 2' I \

o+--1- T l i 1 1 - 7 - ; T'---iT- 0 0 1 2 3 4 5 6 7 ,E 9 10 :: 12 13 14 15 16 17 1E

INSULATION THICKNESS Pm

Figure 4.

Radial distribution of the carbonyl groups mea- sured in the same aged cable than in Figures 2, 3.

interface.

.It was also observed that this density gradient corre- sponds very well to impurity diffusion from the heavily contaminated shields [1,5]. Figure 2 shows a very good

142 Crine et al.: Influence of Insulation Morphology on Electric Properties of Cables

example of significant impurity diffusion from the heavily contaminated shields of a laboratory-aged transmission cable. The concentration gradient near the conductor shield also corresponds to a density gradient (Figure 3) and to an accumulation (Figure 4) of ester groups (see [l, 51 for details). There is also more oxidation near the con- ductor as shown by the concentration gradient of ketone groups (Figure 4). Interestingly, oxidation is associated with polar groups (carbonyls) and organic ions formation 151, both of which could affect the electrical properties of the insulation. I t was speculated elsewhere [l] that aging leads to increased crystallinity away from the shields and also to increased impurity diffusion and oxidation near the shields, as schematically shown in Figure 5.

;g I 12 O X UUY

Insulation thickness +%

Figure 5 Schematic representation of the influence of aging on the density of dry-cured and HMW-PE trans- mission cables. UA = unaged, LA = laboratory aged, FA = field-aged.

AC BREAKDOWN MEASUREMENTS

EVERAL field-aged and one unaged distribution ca- S bles (HMW and XLPE) were broken down a t 60 Hz. The experimental conditions and test procedure were de- scribed in detail elsewhere [6]. The objective of this paper is not to give exact statistical probabilities of breakdown as a function of crystallinity function or impurity con- tent (in fact, nobody is able to do this). Therefore, on- ly minimum and maximum values are reported. The ac breakdown strength values thus obtained are plotted in Figure 6 as a function of the average value of the insu- lation density (taken in the middle of the insulation). A very approximate linear relation is observed for the field- aged cables whereas the unaged one has a much high- er breakdown strength. The observed relation between

1375

ww J NE74

m 77

I I I I I I I I l o ! 0.914 a916 0.918 0.920

p (9 /cm31

Figure 6. Variation of the ac breakdown strength of distri- bution cable (length -5 m) as a function of their average density. See [l] for details.

ac breakdown strength and crystallinity is in good agree- ment with the results of Takahashi et al. [2] obtained with laboratory samples. This does not mean that breakdown strength depends only on polymer density but this is one important contribution.

In the case of transmission cables it was not possible to perform these measurements on full size cables. We per- formed breakdown measurements on ribbons - 100 p m thick peeled off the cables. Results were obtained at 22°C with sphere brass electrodes (12.5 mm dia.) immersed in silicone oil. The voltage was increased with a ramp of 0.5 kV/s. A good example of the radial distribution of breakdown strength obtained with a pair of unaged and laboratory aged dry-cured cables is shown in Figure 7. A small decrease of breakdown strength is visible near the conductor shield of the unaged cable; this decrease is more important near the conductor shield of the lab- oratory aged cable. This corresponds very well with the density, impurity diffusion and oxidation profiles (see also Figure 5). Thus, oxidation and diffusing impurities affect the breakdown strength of actual cables and this is even

IEEE Transactions on Electrical Insulation Vol. 26 No. 1 , February 1991 143

O Z f A 4 1 0 U t f

INSULATION THICKNESS (mm)

170

t ' i ' . p A

Aged 0

. P

b

, Min. 1

b b

lead to voids collapsing [7,8] which further complicate the influence of insulation morphology on cable aging. Nev- ertheless, i t may be concluded that aging induces two competing effects on the electric strength of cables: near the shields, diffusing impurities and oxidation reduce its value whereas it increases in the middle of the insulation due to crystallinity increase.

AC BREAKDOWN vs DENSITY -~ . . -1

180

11 , , , , , , , , A

0 2 4 d A W t I

INSULATION THICKNESS (mm)

Figure 7. Radial distribution of ac breakdown strength of dry-cured unaged and laboratory-aged cables (see their density distributions in Figure 1) .

more important with a low density insulation in which diffusion is easier.

The breakdown strength value in the middle of the in- sulation (i.e. away from the gradient near the shields) varies more or less linearly with its density, as shown in Figure 8 in agreement with [2]. This graph includes also some results obtained with distribution cables. It can be seen that steam-cured insulations with their low density have the lowest breakdown strength. One important mor- phological parameter t o be considered with steam-cured XLPE is the presence of voids with diameter in the mi- crometer range. These voids are likely to be generated during the steam curing and they are often filled with water [l]. The void (which reduces the density) and the water in it will likely reduce the breakdown strength of the polymer, as indeed observed [6]. Thermal aging may

.437

Figure 8. Variation of the ac breakdown strength of ribbons (value taken in the middle of the insulation) as a function of density.

lmulolion lhickness (mm)

Figure 9. Radial distribution of tan6 and E ' measured with the laboratory aged cable whose characteristics are shown in Figures 2-4.

J

I44 Crine et al.: Influence of Insulation Morphology on Electric Properties of Cables

. I I ~ l l l l ~ l l l l ~ l l l l ~ l i l

20 .- 22°C I kHr

15 -- 9 a c s

10 y-

Aged Unwed

I INSULATION THICKNESS

b ) small impurity content large . . 0 5 40 15 20 Insulation lkkness b"m w n

Figure 10. Radial distribution of dielectric losses in unaged and aged dry-cured cable. The loss gradient cor- responds to the density gradient (see Figure 1). -

I

DIELECTRIC LOSSES MEASUREMENTS

HE measured dielectric loss tan6 of a dielectric are the sum of all kind of losses in the dielectric material,

i.e. oxidation by-products, impurities (ionic and dipolar), other dipoles associated with the morphology, etc. For example, i t is well known that the tan6 of PE decreases with increasing crystallinity and i t is also known that oxidation increases tan6 of XLPE [3]. Thus, the measured tan6 is

INSULATION THICKNESS T

Figure 11. Schematic of the influence of crystallinity and im- purities on the dielectric loss of cables aged dif- ferently.

content, oxidation and crystallinity, especially near the conductor shield. tan 6 2 tan 6base + A tan dcryst. +

A tan boxid + A tan 6imP. + . . . where tan6 base is the value of the 'perfect' material i.e. with constant crystallinity, and without oxidation and im- purities.

We have shown schematically in Figure 11 the influence of diffusing impurities and crystallinity change associated with the two different cable aging conditions considered in Figures 9 and 10. A small change of crystallinity near the conductor shield reduces slightly the tan6 contribu- tion due t o crystallinity. At the same time, a large content of diffusing impurities will induce a significant increase of the tan6 contribution due to impurities; this results in

order to verify this assumption we have measured the tan6 and the ( E t ) of cable ribbons at 22-c and 1 kHz with a ~~~~~~l R d i o bridge (model 1616) and with a 3 electrodes ~ ~ l ~ b ~ ~ ~ h cell. Let us consider two very different situations. The Case of a large crystallinity change and a small concentration of diffusing impurities is

a total contribution (Figure 1 l a ) very similar to the One

measured and shown in Figure 10. Conversely, large crys- shown in ~i~~~~ 9. ~~~~h~~ typical tan6 result obtained with a dry-cured cable where impurity diffusion has been important and crystallinity change is small [l] is shown in Figure 10. In the latter case, an important tan6 gradient is seen near the conductor shield. I t therefore seems that indeed tan6 gradient is seen near the conductor shield. I t therefore seems that indeed tan6 varies with impurity

tallinity change and small contents of diffusing impurities lead t o the profile calculated in Figure l l b and measured in (Figure 9).

As expected, the permittivity (measured in the middle of the insulation) decreases with increasing crystallinity, as shown in Figure 12.

IEEE Transactions on Electrical Insulation Vol. 26 No. 1 , February 1991 145

REFERENCES ‘ n t I ’ ” ’ ’ ’ I I I 1

PXL - -Y LE-XL

Unaged .U Aged .LAfiA [l] Diagnostic Techniques for Cable Characterization,

EPRI Report EL-6207, 1989.

[2] T. Takahashi, H. Ohtsuka, H. Takehana and T. Ni- wa, “Study on Improvements to the Dielectric Break- down Strength of Extended Dielectric Cables”, IEEE Trans. PAS, Vol. 104, pp. 1945-1953, 1985.

[3] S. Sapieha and J.-P. Crine, “The influence of Extrud- ed Semiconducting Shields on the Contamination of XLPE Cables”, 1980 Conf. Elec. Insul. Diel. Phe- nom. p. 212, 1980.

2 1 d ” I I ‘ 1 0dp2 d2, A& [4] J.-P. Crine and J. Lanteigne, “Influence of Some Chemical and Mechanical Effects on XLPE Degrada- tion”, IEEE Trans. Elec. Insul. Vol. 19, pp. 220-225, 1984.

?$)IC# 1 0% 096 cm8

Figure 12. Variation of permittivity with insulation density (taken in the middle of the insulation). [5] J.-P. Crine, S. Pdlissou, Y. McNicoll, H. St-Onge,

“A Critical Evaluation of Analytical Techniques for the Characterization of Extruded Dielectric Cables”, IEEE Trans. Elec. Insul., this issue.

CONCLUSIONS

T was shown that insulation morphology (especially I crystallinity), contamination and oxidation affect the ac breakdown strength and the dielectric losses of actu- al cables. Aging may have very different influence on the evolution of the local electrical properties of the insulation due to the competing effects of impurity migration and crystallinity associated with cable aging. These effects are especially important close to the conductor shield and they increase with service aging due to impurity diffusion from the heavily contaminated shield and to increased ox- idation associated with the ‘hot’ conductor. Interestingly, this is the region where breakdown and vented water trees are likely to be initiated. Obviously much more detailed studies are required t o better evaluate these effects.

[6] A. Bartnikas, S. PClissou, H. St-Onge, ‘lac Break- down Characteristics of In-service Aged XLPE Distribution Cables”, IEEE Power Eng. Soc. Conf., Winter Meeting, 1987.

[7] Y. Namiki, H. Shimanuki, F. Aida and M. Morita, “A Study on Microvoids and their Filling in XLPE Insulated Cables”, IEEE Trans. Elec. Insul. Vol. 15, pp. 473-480, 1980.

[8] C. Kats, A. Dima, A. Zidon, M. Esrin, W. Zaengel and B. S. Bernstein, “Emergency Overload Charac- teristics of Extruded Dielectric Cables Operating a t 13OOC and Above”, Proc. of 1984 Transm. and Dis- trib. Conf. Paper 332-2, 1984.

This manuscript is based on a paper given a t the 2nd Interna- tional Conference on Properties and Applications of Dielectric Materials, Beijing, China, 12-1 6 September 1988.

ACKNOWLEDGMENTS

Manuscript was received on 12 Mar 1990, in final form 9 Nov 1990. This work has been partially supported by EPRI

(RP7897-2) and we thank R. S a m and B. S. Bernstein for their support and comments.