8/12/2019 DK2871_CH06

1/14

CHAPTER 6ELECTRICAL PROPERTIES OFINSULATlNG MATERIALSBruce S. Bernstein1. INTRODUCTIONElectrical properties of interest for insulation materials can be classified into twomajor categories:

Those of significance at low voltage operatings t ressesThose ofimport nce t high voltage operating stresses

At low stresses, the properties of interest relate to dielectric constant, powerfactor, and conductivity resistivity). Dielectric constant represents the ability ofthe insulation to hold charge. Power factor representsa measure of the amountof energy lost as heat rather than transmitted as electrical energy. A gooddielectric (insulation) material is one that holds little charge (low dielectricconstant) and has very low losses (low power factor). Polyolefins representexamples of polymers that possess excellent combinations of these properties.This s discussed in depth in Chapter 5 At high stresses greater than operating s t ress the characteristic ofimportance is dielectricstrength. Here, the insulation must be resistant to partialdiscbarges (decomposition of air in voids or microvoids within the insulation).Also of interest is the inherent ability of the polymeric insulation material toresist decompositionunder voltage stress. Unfortunately, the measured dielectricstrength is not a constant but has a variable value depending upon how themeasurement is performed. This will be discussed later in this chapter. In anyevent, the dielectric strength must be high* for the insulation to be functional.This chapter will review factors that influence electrical properties at low andhighvoltage stresses.

87Copyright 1999 by Marcel Dekker, Inc.

http://dk2871_ch05.pdf/http://dk2871_ch05.pdf/http://dk2871_ch05.pdf/

8/12/2019 DK2871_CH06

2/14

2. STRUCTUREPROPERTY RELATIONSHIPSThe electrical properties of an insulation materials are controlled by theirchemical structure. Chapter 5 reviewed the inherent chemical stru ture ofpolyolefins, and described how the structure influences physicochemicaiproperties. In this chapter, we sh ll review how these factors influence theelectrical properties. The emphasisshallbe on polyolefins.Low stress electrical properties are determined by the polar nature of thepolymer chains and their degree of polarity. Polyethylene, composed of carbonand hydrogen or methylene chains, is non-polar in nature and has lowconductivity. If a polar component, such as a carbonyl, is on the chain, thepolymer chain now becomes more polar and the characteristics hat lead to lowconductivity are diminished. Ethylene copolymers wth propylene retain theirnon-polar nature since the propylene moiety is as non-polar as is the ethylenemoiety.When a polyolefin is subjected to an electrical field, the polymer chains have atendency to become polarized. Figure 6-1 shows what happens whena polymeris stressed between electrodes, wt different polarities resulting.Figure 6-2shows how the polymer insulation material responds. There is a tendency for thepositive charges on he polymer to move toward the negative electrode, and forthe negative charges on the polymer to move toward the positive electrode,hence pulling the polymer in two directions. This is a g a d escription, anddoes not take into account the chemicalstructure, which isdiscussed later.Figure6-1Polarization of a Polymer Subjected to an Electric Field

INo Field Field AppliedPolymer Becomes Polarized

Schematic description of a polymer subjected to electric field; polymer becomespolarized.

88Copyright 1999 by Marcel Dekker, Inc.

http://dk2871_ch05.pdf/http://dk2871_ch05.pdf/http://dk2871_ch05.pdf/http://dk2871_ch05.pdf/http://dk2871_ch05.pdf/http://dk2871_ch05.pdf/

8/12/2019 DK2871_CH06

3/14

Figure 6-2Charge Migration on Polymer Cbains Subjected to Electr ic Field

Electrode Polymer Electrode Electrode Polymer ElectrodePositive) Negative)

No Field Field Applied

Insulation response to electric field application. Positive charges on polymerchain migrate toward the cathode and negative charges migrate toward theanode.Where do these charges come from? After all, we have described the polyolefinsas being comprised of carbon and hydrogen, and as not being polar compared tosay the polyamides or ethylene copolymers possessing carbonyl or carbo;\?;lategroups. It can be noted that such description is ideal in nature. While beingtechnically correct for a pure polyoolefin, in the realworld there are always smallamounts of such polar materialspresent Thiswill be discussed later.Figure 6-3 shows what may happen o a polymer insulation material that haspolar groups on the side branches, rather th n on the main polymer chain. Notethat in this idealized description of the folded chain, the m in chain does notundergo any movement under voltage s t r e s s The side chains, which were oncerandom, are now aligned toward the electrodes.Figure 6-4 shows a morerealistic coiled polymer chain with polar branches. Note how the alignmenttoward the positive and negative electrodeshastakenplace.

89Copyright 1999 by Marcel Dekker, Inc.

8/12/2019 DK2871_CH06

4/14

Figure 6 3Schematic D escription of Orientation of Polar Functionality on PolymerSide Chains Subjected to Electric Field

No Voltage Voltage Stress AppliedUnder voltage stress, a polar chain orients toward the cathode or anode.depending upon the charge it possess.The nan-polar chain does not migrate.Figure 6-4Polarizationof Side Chains Depicted on a Coiled Polymer

, Io Voltage Field Field ApptiedPolymer BecomesPolarized

A polymer is typically coiled, as shown here. The positive charges on a polymerare attracted to the cathode. The negative charges are attracted toward the anode.The movement of these charged regionscausesmotion of the entire side chain.In Figure 6-5 we show what happens to the main chain. Prior to this we hadconsidered what happened to the relatively short branches. However, the entiremain chain m a y undergo motion also, assuming it possesses functional groupsth t respond to the voltage stress. The figure shows that entire chain segmentsmay move and rotate, in accordance with the field

90Copyright 1999 by Marcel Dekker, Inc.

8/12/2019 DK2871_CH06

5/14

8/12/2019 DK2871_CH06

6/14

be some oxidized functional groups on the polymer chains. These are importantpoints to keep in mind when reviewing the polymer insulation response tofrequency.3.0 DIELECTRIC CONSTANT AND POWER FACTORDifferent regions of the polymer chains will be sensitive and respond differentlyto voltage stress. This phenomena is intimately related to the ftequency.Different hnc tional groups will be sensitive to different frequencies. When theproper frequency-functional group combination occurs, the chain portion willrespond by moving, e.g ., rotating. Since this phenomenon is frequencydependent, one might expect that different responses will result from differentfunctional group-frequency combinations. This is exactly what occurs. Referringto the top curve in Figure 6-6, we can see that at low frequencies, when s t r e s s isapplied, the polar region-dipoles-can respond and accept the charge, and alignas described above. The dielectric constant is relatively high under theseconditions. As the fiequency increases, no change occurs in this effect will occuras long as the dipoles can respond. At some point as the frequency continues toincrease, the chains will have difficulty responding as fast as the field i schanging. When the fiequency change is occurring at so rapid a rate that norotation can occur, the charge canno t be held and the dielectric constan t will belowered.Figure 6-6Dielectric C onstant and Power Factor as a Function ofFrequency

I I I Il og

log YJUpper portion of Figure 6-6 depicts the change in dielectric constant withfrequency. The lower portion of the figure depicts the change in power factorwith frequency.

92Copyright 1999 by Marcel Dekker, Inc.

8/12/2019 DK2871_CH06

7/14

For a polymer like polyethylene, with very small amounts of polar functionalitythe dielectric constant is always low (compared to a more polar polymer such asa polyamide [Nylon for example]). However, oxidized regions will respondmore readily due to their more polar nature, The reason for the change indielectric constant with fresuency is clear. It should also be noted that otherparameters affect this property; e.g., temperature. In essence, any change thata f k c t s motion of the polymer chain will affect the dielectric constant.The point where the polymer ch in segments undergo change in rate of rotationis of special interest. The lower curve of Figure 6-6, focusing on losses (e.g.,power factor), shows a peak at this point. In considexing power factor, the sameexplanation applies; changes are affected by frequency and specific polymernature. At low frequencies, the dipoles on the polymer chains follow variationsin the ac field, and thecurrentandvoltage are out of phase; hence the losses arelow. At very high fkquencies as noted above, the dipoles cannot move rapidlyenough to respond,and hence the losses are low here also. But where the changeis taking place, the losses are greatest. This can be visualized by thinking interms of motion causing the energy to be mechanical rather thanelectrical innature. t is common o refer to the dielectricconstant and power factor at 50 or60Hertz, and at 1,000hem.In relating the information shown in Figure 6-6 to the earlier figures, it is to benoted that the polar functionality can be due to motion of main chains orbranches. Where the oxidized groups are the same, as in carbonyl, one couldexpect that the chains (ideally) to respond the same way at the same frequency.But what happens if there are different functional groups present such as acadmnyl, carboxyl, or even amide or imide functionality? Also how does themain chain nature affect all this? The answer is that these factors are quitesignificant. Different functional groups will respond differently at the samefrequency, and the main chain can hinder motion due to its viscoelastic nature.If the dipole is rigidly attached on the polymer backbone, then main chainmotion is going o be involved. If the dipole is on a branch, it c n be consideredto be flexibly attached,and the rate of motion of the branch wll be expected todiffer from the main chain, even if the functional group is the same. The endresult of all of this is a phenomenon called dispersion. Here the chains move atWerent rates at any single fresuency and temperature. They may exhibit achange Over a broad region ratherthan a sharp, localizedregion as the frequencyand temperatureis changed slightly.For purposes of understanding power cable insulation response, the maininterest is, of course at 50 or 60hertz.Also our interest is in what is intended tobe relatively non-polar systems. It is necessary to remember that no system isperfect and there will be variations in degrees of polarity not only from oneinsulation material to another, and not only from one grade of the same material

93Copyright 1999 by Marcel Dekker, Inc.

8/12/2019 DK2871_CH06

8/14

to another, but perhaps also form one batch of supposedly identical material toanother. Much depends upon the processing control parameters duringextrusion.The literature reports dielectriclossesof many M er e n t types of polyolefins as afunction of temperahue, at controlled frequencies. Hence, it is known thatconventional low density polyethylene undergoes losses at various Me r e n ttemperatures. In addition, antioxidants, and antioxidant degradation by products,low molecular weight molecules, will also respond, and this complicatesinterpretation. With conventional crosslinked polyethylene, the situation is evenmore complex as there are peroxide residues and crosslinking agent by-products.These low molecular weight organic molecules, acetophenone, dimethyl benzylalcohol, alpha methyl styrene, and smaller quantities of other compounds wllgradually migrate out of the insulation over time. Hence interpretation of datarequires not only knowledge of the system, but some degree of caution isprudent. In addition to all of this, if there are foreign contaminants present, it ispossible that they also c n influence the m e a d ielectric constant and powerfactor.The dielectric constant of polyethylene is dependent upon the temperature andfresuency of testing. At constant temperature, it is reduced slightly as thefresuency increases; at constant frequency, it increases with temperature.4. DIELECTRIC STRENGTEThe dielectric strength of an insulation material can be defined as the limitingvoltage stress beyond which the dielectric c nno longer maintain its integrity.The applied s t ress causes the insulation to fail; a discharge occurswhich causesthe insulation to rupture.Once that happens, it can no longer serve its intendedrole. Unfortunately, the dielectric strength is not an absolute number; the valueobtained when dielectric strength is measured depends on many factors, not theleast of which is how the test is performed. Therefore, it is necessary to reviewthe issues involved, so that the value and the limitations of the term dielectricstrength are well understood.The dielectric strength i s usually expressed in stressper unit thickness--volts permil or kV per mm. For full size cable, it is common to merely report the kV atwhich the cable has failed. Hence if a 175 mil wall cable fails at 52.5 kV or52 500 olts), the dielectric strength can also be expressed as 300 V/mil.The most obvious value of dielectric strength is called the intrinsic strength.This is defined by the characteristics of the material itself in its pure and defect-freestate, measured under test conditions that produce breakdown at the highestpossible voltage stress. In practice, this is never achieved experimentally. One

94Copyright 1999 by Marcel Dekker, Inc.

8/12/2019 DK2871_CH06

9/14

reason asnoted above, is the diEculty in attaining a defect-free pure insulationspecimen. The closest one can come is on measurement of very thn carefullyprepared films with appropriate electrodes. (The thinner the film, the less thechance for a d e f d to exist.) Under these ideal conditions, the insulation itselfwould fail due to its inherentproperties (bond strength rupture).It is mom likely is that hilure will occur uuder discharge conditions; hem gas(e.g., air) present in small voids in the insulation,present due to processingcharacteristics,will undergo decomposition. Air is the most likely gas presentfor polyethylene and crosslinked polyethylene (in contrast to vapors ofcrosslinking by products). Its intrinsc dielectric strength is significantly lessthanthat of polyethylene. Under these conditions, the discharges that take placein these sm a l l void@) eads to erosion of the insulation surface in contact withthe air. This in turn leads to gradual decomposition of the insulation andeventual failure. The decomposition of the air in the voids o urs at voltagestresses much lower than the inherent strength of the polyethylene itself, Forexample, the dielectricstrengthof one mil thickfilmof polyethylene measuredunder identicaI conditions to a layer of air (atmospheric pressure), gives adielectric strengtb value 200 times greater. Polyethylene give value of about16,500 volts per mil, while that of air is about 79.The dielectric strength of airincreases with pressure that of polyethylene does not change), and this concepthas commercial impact; however the degree of improvement is small. Byincreasing the pressure by a factor of 6, the dielectric strength increases by afactor of about 5 stillwell below that ofthe polymerfilmWhen focusing on emded cable insulation, we are now concerned withrelatively thick sections; 175 to 425 mil walls for distribution cables, and eventhicker walls for transmission cables. Discharges that OCCUT in these practicalsystems may not lead to immediate failure. It is possible that the discharge willcause rupture of a portion of the wall, and then cease. This could be related tothe energy of the discharge, the size of the adjacent void, and, of course thenature of the insulation material.When this occ~rs,we will developa blackenedneedle-shaped series of defects, sometimes resembling a tree limb; these arecalled electrical trees.Dischargesmay occur repetitively, and hence the tree willappear to grow In time the bee will bridge the entire insulation w ll andcause failure. Discharges may also occur on the surface of the insulation,particularly if there ispoor adhesion between the insulation and shield layers.Another mechanism of failure is known as thermal breakdown. This occurswhen the insulation tempemure starts to increase as a result of agingphenomena under operating stress. Under voltage stress, some insulationsystems will start to generate heat, due to losses. If the rate of heating exceedsthe rate of cooling (that normally occus by thermal tmsfer) then thermalrunaway occurs, and the insulation fails by essentially, thermally induced

95Copyright 1999 by Marcel Dekker, Inc.

8/12/2019 DK2871_CH06

10/14

degradation. Several points should be kept in mind here:1) The heat transfer capability of polyolefins is low, and heat dissi-pation is not normally rapid2) These events may occur in the presence or absence of discharges

(3) The presence of inorganic fillers contributes to increasing thedielectric losses, and may exacerba te the situation. Also, some organicadditives in the insulation may also lead to increasing the dielectriclosses/ Finally, it should be noted that thermal breakdown ofpoylolephins is a very well-studied area.

Although not a direct cause of failure, mention should be made of water treeing;water trees lead to a reduction in dielectric strength , but are not a direct cause offailure. These trees have a different shape for electrical trees, and also havedifferent cause. The differences are outlined below.

WATER TREES ELECTRICAL TREESWater required Water not requiredFan or bush shapedGrow for yearsMicrovoids connected by tracks

Needle o r spindle shapedFailure shortly after formationCarbonized regionsWater trees grow under low normal) operating stress, do not require thepresence of small voids, and lead to a reduction in dielectric strength.Laboratory studies have shown that such trees can penetrate virtually the entireinsulation wall yet not lead immediately to failure. As the chart shows, thechannels or tracks that comprise water and electrical trees d iffer.A C breakdown strength is commonly performed on f i l l size cables as an aid incharacterization. For full size cables , it is common to perform many such tests oflong lengths of cables e.g., 25 to 30 feet) and plot the data on WeibulI or Lognormal curves. This is done as the data always has some variation. A goodexample is data developed on a project for the Electric Pow er Research Institute(EPRI).

96Copyright 1999 by Marcel Dekker, Inc.

8/12/2019 DK2871_CH06

11/14

Figure 6-7AC B m a k d m Strength of15 kV XLPE Insdated Cabk as a Function ofPosition on Reel that Contained 5,000 F a t of Cabk and Total ErCrosionRun wm 50,000 Feet

AC Breakdown in volts per mil140012001000800600400200 0 80 160 240 320 400 480

Position on Cable Run in FeetInFigure 6-7, it is Seen that the dielectric s a ~ n g t h f full size c a b b varies h na low of about 600 V/mil to a m ximum of about 1,300 Vlmil. Thisdemonstrates that dthough the cable was manltEacturedin presumably the w ema~e rthis cable was tested from the same ex usionnm and the same reel),some variation is inevitable. This is appannt from the ac breakdown strengthmeasurementandisthereasonthatsmtisticaievaluationofthedataobtainedisa necessity. From what has been noted above it is likely that theseVariations aredue to inevitable imperfections that result during process@ Figure 6 7demonstrates the variation in measured ac brrakdown saength of cmslinkedpolyethylene insulated cable. Sample lengths tested wtre from the sameproductionrn ndfrom the same reel. Variationssuch as hese an common andare the reaSOn for employing statistical analysis of data such as WeibulldistributionThe data shown fium the EPRI project was obtained at a five minute step risetime. If the t ime interval between the steps is i n d e.g., from 5 to 10minutes , the apparent ac breakdown strength decreases. If the t ime intendbenveen steps is increased again to say 30 minutes the appeut dielearic

97Copyright 1999 by Marcel Dekker, Inc.

8/12/2019 DK2871_CH06

12/14

strength is reduced even more. In other words, the apparent dielectric strengththat one obtains in performing a test increases as the stress is applied morerapidly. [This is analogous to what happens during a tensile strength test forpolyethylene; the apparent tensile strength increases as he stress is applied morerapidly]. Therefore, the meaning of an ac breakdown strength value is relevantonly if the manner in which the test was performed is known. In comparing acdielectric strength values for different insulation materials, the test shouldalways be performed in the same manner. This holds true whether one iscomparing Werent grades of the same material W eren t grades ofpolyethylene) or W e ren t insulation materials polyethylene versuspolypropylene), for example.The testing of thin films or slabs of insulation materials is performed in thelaboratory and the opportunity to control the local environment during testing i spresent. This should be done and should be reported. Since relatively smallspecimens are involved compared to fit11 size cables), a large number areusually tested to overcome the inherent variability in results, as noted above.When working with small samples, the opportunity to control the localenvironment during testing is greater and reproducibility may be e n h a n d .Hence the following variables are to be controlled so that the informationobtained represents a true representation of the statistical distribution inhomogenieties for the m aterial under study.

specimen thicknesstemperatureelectrode shape and sizeThe reasons for controlling the thickness have been noted. This is especially ofincreasing importance as the thickness is reduced. Temperame control is vital,as he dielectric strength is related to the temperature of the specimen at the timebreakdown occurs. Clearly, when working with small samples, the opportunityto generate experimental data at controlled uniform temperatures [such as bytesting in a controlled environment room, or in an oven] is present.The electrode shape and size represents a significant parameter for small sampletesting. The most common electrodesare Rogowski t ypes, where the electrode ism e d and inserted into the polymer slab; this provides a uniform stress gradientand enhances the oppoxtunity for obtaining meaningfid information. If theelectrode-polymer interface is sharp instead of rounded) the voltage stress willbe enhanced at this location.The failure of the test specimen will be induced atthis location. Should that happen, the dielectricstrength measured will be relatedmore to the manner in which the test was performed inducing a high localizedstress) rather than elated to the properties of the insulation itself.

98Copyright 1999 by Marcel Dekker, Inc.

8/12/2019 DK2871_CH06

13/14

Needle tests are also performed, where a sharp, but controlled radius ofcurvature exists at the needle tip, and the latter isinserted into the specimenpartway to the ground plane, Voltage stress is applied and the dielectric strength ismeasured; his approach has been used to determine the influence of additives,designed to incmse the breakdown strength, and aid in developing superiorinsulation materials. A detailed description of typical amngements of electrodesthat m y be used for dielectric strength testing of thin films is provided byMathes in the references.The fresuency of me as me nt may be readily varied in thin film studies, muchmore easily thanfor full s i ze cables. While most testing is performed at 60 hertz,testing has also been performed at fresuencies ranging to 1,OOO hertz. Again therate of riseof the field is vitally importanz and can readily be controlled.The reasons br controlling the thickness have been noted above. This isespecially critical when working with thin samples. Temperaturecontrol is alsovital, as the dielectric strength is related to the temperature of the specimen atthe time breakdown occurs. Clearly, when working with small samples, theopportunity to generate experimentaldata at oontrolled,d o r m emperatures ispresent. For instance, do the testing in a controlled environmentroom or oven.The electrode shapeand size represents a significant parameter for small sampletesting. The most common electrodesare Rogowski types where the electrode iscurved and inserted into the polymer slab. This provides a uniform s t ressgradientand enhances the opportunity for obtaining meaningfulinformation.If the electrode-polymer interface is sharp instead of rounded), the voltagest ress will be enhanced at this location. Failure of the test specimen will beinduced at this location. Should that happen he dielectric strength measuredwill be relatedmore to the manner in which the test was performed (inducing ahigh localizeds t ress rather than related to the properties of the insulation itself.Needle tests are also performed where a sharp generally controlled radius ofcurvature exists at the needle tip. This needle is then inserted into the specimenpart way to the ground plane. Voltage s t ress is applied and the dielectricstrengthis measured. This approach has been used to detennine the influence ofadditives, designed to increase the breakdown strength and an aid in developingsuperiormaterials. A detailed description of typical anangements of electrodesisprovided by Mathes in the references. The frequency of measurement may bereadily varied in film studies much more easily thanfor full size cables. Whilemost testing is performed at 60 hertz, testing has also been performed atfrequencies ranging to 1,OOO hertz. Again, the rate of rise of the field is vitallyimportant andcan readilybe controlled.

99Copyright 1999 by Marcel Dekker, Inc.

8/12/2019 DK2871_CH06

14/14

5. SUMMARYThe chem ical structure of the po lymeric insulation determines the magnitude ofthe d ielectric constant and power factor. These two properties are significant atoperating stress and generally considered to be low. Polyolefins such as poly-ethylene or crosslinked polyethylene have low dielectric constants and low pow-er factors. Low levels of oxidation, generally resulting from processing the poly-mer, lead to slight increases in these properties. Higher than normal operatingstresses are used to determine the dielectric strength of an insulating material.The manner in which the test is designed and performed can influence the re-sults. Statistical evaluation of the dielectric strength data is required. Failuremechanisms are briefly reviewed and the differences between water and elec-trical trees are noted.6. REFERENCES[6-11 L. A . Dissado and J. C Fothergill, Electrical Degredation and Break-down in Polymers, G. C. Stevens, Editor, Peter Peregrinus Ltd., 1992.[6-21 Ken M athes, Electrical Insulating Materials.[6-31 M. L. Miller, The Structure of Polymers, Reinhold Book Corpo ration,SPE P olymer Sc ience and E ngineering Series, Chap ters 1 2, 3, 10, 13, 1966.