Extended analysis versus frequency of partial discharges phenomena, in support of quality assessment of insulating systems Romeo C. Ciobanu, Cristina Schreiner, Ramona Burlacu, Cristina Bratescu Technical University Iasi Dept. of Electrical Measurements and Materials Bdul D. Mangeron no. 53-55, IASI, 700050, Romania Tel.: + 40 232 211352 Fax: + 40 232 237627 E-mail: rciobanu@ee.tuiasi.ro Abstract The broadband Partial Discharge (PD) measurement technique may be considered more accurate than a simple breakdown test and admitted as a complementary method for characterizing the quality of porous/fibrous and composite insulating materials. It should be taken into account especially at higher frequencies, when the risk of breakdown occurrence becomes greater in time, due to the local thermal aspects. Keywords: Partial discharges, composite insulating systems, cable paper technology 1. Introduction Partial discharges usually occur, if in gas filled cavities within solid insulation or in air gaps between several layers within an insulation construction, respectively between insulating material and electrodes, the breakdown voltage of the enclosed gas (mostly air) is exceeded. In case of capacitive voltage distribution for a.c. voltages the cavities are usually stressed much higher and also the PD-impulse rate is by magnitudes higher. As regards the standardisation of PD-testing, we need to accept PD-testing as a general recognised tool to verify the integrity of high-voltage insulation systems. These specifications have been introduced in the international safety standard for insulation coordination of low-voltage equipment, as far as solid insulation is concerned [1]. For thin insulation systems (< 1 mm) and especially for safety-related applications, the verification of the non-existence of PD or the capability to withstand PD permanently should always be required. By performing PD-testing both requirements for minimum thickness through insulation and excessive amplitudes of the test voltage and duration of the test are substituted. There is no physical background for the values of minimum thickness through insulation, which are currently required and which vary between 0.4 and 2 mm. It is difficult to understand how such different thickness values can provide the same required degree of safety. The use of very high test voltages without measuring PDs always implies the risk of degradation of the solid insulation. Such tests even widely accepted - can only show the important defects within the insulation system and can be acceptable in case of type testing, but they are not appropriate for sample or routine testing. The PD-testing is not necessarily performed with high expenditure. This may be at very high voltage levels, when the achievement of low interference levels is a real problem. But for test voltages less than 10 kv and according to the simplified test procedures, the PD-test is comparable to the traditional power-frequency highvoltage test. Simple screening measures are sufficient to meet the required interference level and PD-test apparatus with inherent interference suppression are widely available nowadays. 2. PD-testing at higher frequency Requirements and guidance for PD-testing are contained in e.g. [2, 3]. In high-voltage engineering the requirement of PD-testing is recognised in general, as the existence of high field stress is obvious. However, due to very small insulating distances in recent low-voltage 4-307
equipment similar or even higher fields can occur. This is usually the case if increased stresses due to overvoltages or dielectric testing are applied. In the past, this aspect was only taken into account when very high reliability was required [2, 4]. But recent technologies with decreasing insulating distance require a reconsideration of this situation. As a lower physical limit for the occurrence of PD, a peak voltage of 300 V may be considered. According to practical experience PD can occur when peak voltages in excess of 500 V are applied. 1 Power generator 10 Hz - 400 khz 2 Wide band high-voltage transformer 3 Test specimen 4 High pass filter 5 High-voltage probe 6 Screened cabinet 7 High speed digital storage oscilloscope 8 Decoupling amplifier 9 PD-measuring instrument (narrow band) 10 Digital voltmeter 11 Analogue oscilloscope 12 Control computer Fig. 1. PD-test circuit for higher frequency voltage tests. With respect to any kind of high-voltage testing at frequencies which are much higher than the power frequency, the availability of appropriate test voltage sources with adjustable frequency is a fundamental problem. The PD-measurements with higher frequency test voltage, which are reported here, have been performed with the test circuit shown in figure 1. This circuit is based on a high-frequency resonance transformer (2), which is fed by a highfrequency power generator (1) [5]. A possible alternative is a high-frequency power oscillator, which allows the generation of significantly higher frequencies [6]. For measuring the PDintensity, a digital storage oscilloscope in combination with a rejection filter with steep characteristic in order to suppress the high-frequency test voltage was basically used. 4-308
Additional hardware and software facilities were created in order outline the measured quantities: PD pulse peak value of inception and extinction voltage, apparent charge of individual PD pulses, instantaneous values of test voltage and frequency when PD pulse occurs, other statistical data related to PD distribution versus phase or charge value. 3. Experimental investigations A wide literature describes the negative role of ions upon dielectric properties of hygroscopic materials, especially upon breakdown strength. However it proved to be no simple task to appoint a fair mineral purity limit of a certain insulator, especially under the circumstances when divergent opinions are occurring, regarding the real dependence between this limit (imposed by producers, on economical reasons) and the values of some dielectric properties (strictly imposed by the insulation users, on service or reliability reasons). Our study relates the case of insulation based on cellulose pulps for electrical purposes. The investigations were purchased on laboratory made samples of paper for cable applications, from demineralised - unbleached softwood sulphate - pulps, industrially beaten till a refining degree of 40 SR. The demineralisation process followed and presumed a special industrial treatment with HCL-acid (an equivalent addition of 80 kg acid / tone of absolutely dried cellulose) under controlled demineralisation duration (till a maximum admitted value - 2 hours), as widely described in [7]. All paper samples were finally formed and adjusted to 70 g/m² (0.12 mm thin). The results were correlated directly with the mineral content in paper, c ash [%]. A direct comparison of PD characteristics with the standard breakdown test was provided. All measurement results are presented as RMS values. It was assumed that the analysis of broadband PD phenomena versus mineral content of papers could offer new information in order to improve the theory of cellulose dielectric failure. The need for such studies at fundamental level is without doubt nowadays. 800 600 [ V ] 400 Ud - 1% Ud - 0.42% 200 Ui - 1% Ui - 0.42% 50 150 350 1 k 2 k 3 k 3.5 k 10 k 30 k 35 k 45 k 60 k 110 k Fig. 2. Comparative evolution of breakdown and PD inception voltages versus frequency, for 1% and respectively 0.42% mineral content in paper. U d - breakdown voltage / U i - PD inception voltage / U e - PD extinction voltage. After testing the PD-characteristics, it was found that the most significant variation versus mineral content was emphasised mainly by the inception and extinction voltages, U i and U e, in the whole frequency range. That is why only the evolution of these characteristics is presented herein. On the other hand, when comparing the PD and breakdown processes, there were found appreciable differences between the evolution versus frequency of breakdown and PD inception voltages. This observation occurred reproducible for all types of analysed 4-309
samples, having different mineral content. But in order to express more synthetically the influence of mineral content, only two reference values are presented below, respectively: 0.42 %, considered the optimum value in the whole frequency range, and 1%, the standard reference content for common papers for electrical purposes. Figure 2 shows the continuous decrease of the breakdown voltage, U d, with the increase of frequency homologue to the one described by literature, but also the peculiar unexpected evolution of U i, different from U d. This phenomenon appointed herein is considered new, emphasised for the first time by the authors. This aspect was proven to be reproducible in a more general matter, for a wide range of dielectric structures and technologies analysed elsewhere by authors, even under different stress parameters or related to other physical or chemical features. 4. Analysis of experimental data At industrial frequencies, the PD phenomena appear clearly for lower voltages, comparing to the homologue breakdown value and this aspect could be a good explanation for the premature ageing of dielectric material in certain applications of electrical drives. The decrease becomes even more important in the case of higher mineral content (herein 1%). Basically, at over 10 khz the evolution versus frequency of both PD and breakdown phenomena look concordant as described in literature. In this domain the mineral content seems to influence only the PD inception voltage, comparing to the breakdown behaviour. That is why new studies of PD phenomena are needed in future, in order to understand more accurately the intimate pre-breakdown processes and failure of dielectric structures. 25 20 15 10 5 [ % ] 50 Hz 150 Hz 1 khz 45 khz 110 khz 0-5 0,2 0,4 0,6 cash [ % ] 0,8 1 1,2 Fig. 3. PD - Inception voltage evolution versus mineral content in paper, for different frequencies. 25 20 15 10 5 0-5 -10 [ % ] 50 Hz 150 Hz 1 khz 45 khz 110 khz 0,2 0,4 0,6 c ash [ % ] 0,8 1 1,2 Fig. 4. PD - Extinction voltage evolution versus mineral content in paper, for different frequencies. In order to account for the demineralisation process efficiency via broadband PD analysis, in figures 3 and 4 the evolution versus mineral content of the relative variation of the PD inception and extinction voltages is presented. It was noticed that a significant variation of these characteristics occurred within a relatively small domain of mineral content, with an optimum of about 0.42 %. The increase of U i and U e - values appears obviously due to the 4-310
diminishing of ions content in paper, reducing the effect of PD phenomena. These results are according to similar observations obtained for other insulations, as thin films or cast resins. Once the mineral content becomes less than 0.4%, the positive effect of demineralisation process is progressively lost, especially at higher frequencies. This expected behaviour could be explained by the presence of the residual Cl - ion and by the chemical ageing process consequent to the longer demineralisation process duration, needed to reach much lower mineral content values. In our study, the optimum value of mineral content seen as a technological demineralisation limit was found 0.42 %, instead of 0,35 %, recommended by literature by analysing the breakdown characteristic alone, [8]. It seems that by applying short-term high electrical stresses, the case of breakdown strength, a higher mineral purity should be needed to reach some theoretical recommended values. But exactly the chemical damages incurred by the obtaining of this purity limit favour the occurrence of PD phenomena during a long term exploitation process, leading finally to a reduced lifetime of insulation. This is a clear example of doubt in relation with breakdown test purpose and validity related to samples routine testing. We finally estimated that the optimum value of 0.42 % can be considered more accurate and closer to the reality of long term in-service stress of papers for electrical applications. This value, if admitted as valid for cable paper, would provide significant technological and financial advantages for paper producers, reflected in a diminished consume of reactives and /or in a reduced duration of demineralisation. 5. Conclusion The appreciable differences between the evolution versus frequency of the PD and breakdown phenomena, aspect reproducible for many types of insulating specimens, suggests that the breakdown analysis alone proves to have limits in accuracy and results interpretation. The PD occurrence is clearly influenced by the mineral content, and an optimum value of 0.42 % was appointed for cable paper, for the whole frequency range. Such values, validated via PD analysis, can provide significant technological advantages for the producers. The extended analysis versus frequency of PD phenomena is considered a valuable complementary method in quality assessment of insulating technologies, and even for sample testing, being more close to the real in service behaviour and ageing mechanism of insulation. References 1. IEC 60664-1/1992. Insulation coordination for equipment within low-voltage systems Part 1: Principles, requirements and tests. 2. IEC 60270/1981. Partial discharge measurements. 3. IEC 61180. High-voltage test techniques for low-voltage equipment Part 1/1992: Definitions, test and procedure requirements; Part 2/1994: Test equipment. 4. IEC 60664-4/1997. Insulation coordination within low-voltage systems Part 4: Considerations of high-frequency voltage stress. 5. W. Pfeiffer. Dielectric Testing of Solid Insulation with Respect to Insulation Coordination of Low-Voltage Equipment. IEEE Transactions on Electrical Insulation. 2001, vol. 17, No. 3, pp. 34-47. 6. W. Pfeiffer. High frequency voltage stress of insulation - methods of testing. IEEE Transactions on Electrical Insulation. 1991, vol. 26, pp. 239-246. 7. R. Ciobanu, W. Pfeiffer. Influence of mineral content on the behavior of electrical papers under high frequency partial discharges. European Trans. on Elec. Power Eng. ETEP. 1999, vol. 9, No. 3, pp. 193-197. 8. R. Ciobanu, W. Pfeiffer, E. Poppel. A theoretical approach to the P.D. phenomena in composite dielectrics. Cellulose Chemistry and Technology. 1997, No. 5, pp. 624-627. 4-311