B6.6

9th International Conference on Insulated Power Cables

B6.6

Loss of dielectric strength of polymers due to high-frequency voltages in HVDC applications Matthias Birle, Carsten Leu, Technische Universität Ilmenau, Research Unit High-Voltage Technologies, Ilmenau, Germany, [email protected], [email protected]

ABSTRACT This paper deals with the thermal-electrical stress of polymers in HVDC applications due to mixed voltages with high frequency content. Based on measured breakdown voltages of PVC in a homogeneous field, the DSC-method (Differential Scanning Calorimetry) is used to analyse the specimen with regards to local overheating and influence of the mixed voltage. Further current, voltage and temperature measuring signals of different polymers are presented to analyse the process immediately before loss of dielectric strength of the specimen. Additionally the alteration of electrical material data during the test is calculated based on the measured quantities.

KEYWORDS Mixed voltages, high-frequency high-voltage, HVDC, polymers, dielectric heating, breakdown, PVC, PMMA, POM, DSC, differential scanning calorimetry

INTRODUCTION In conjunction with the changing structure of our energy grid to renewable energy production, innovative converter topologies and HVDC technologies are used to transmit and convert electrical energy in an optimized manner. Thus mixed voltages, such as high direct or power frequency voltages superimposed by repetitive transients or square voltage forms, are produced by the functional principle of the converter and power electronic valves switching. Insulation materials such as polymers have to withstand these different types of mixed voltages to ensure a reliable operation of electrical apparatus in our power grid.

METHODOLODGY OF MIXED VOLTAGE TESTING Mixed voltage forms, which typically appear in praxis, are enormous diverse and depends on the converter topology, the load case or damping of high frequency components. Therefore it is not useful to test insulation materials with any voltage form, but with well-defined mixed voltage forms. The aim is to understand basic effects and the influence of superimposed dielectric and conductive fields, amplitudes and frequencies. The synthesis of mixed voltage as a test voltage depends on the impact of the mixed voltage, which is in the focus of interest. Here, a well-defined heat source density and sinusoidal voltage forms are necessary. Hence at the used mixed voltage forms the characteristic form of the subordinated voltage is maintained, because it is the actual electrical field stress. The energy input due to dielectric heating is realized by a superimposed sinusoidal high frequency voltage. Therefore high voltage circuits are realized with the use of a high-frequency high-voltage generator, based on the resonance principle and a SF6 – polymer-insulation system [5, 6]. Particular attention should be paid on the correct measurement of the mixed voltage. Figure 1 shows the used principle to synthesize test voltages for dielectric heating investigations from mixed voltages which occur in praxis.

The effect of high frequency voltages on insulation materials or insulation systems is different to power frequency voltages respectively direct voltages. The intensity of effects which leads to a decrease in electrical strength in a long-term view is usually higher. These effects are • the heating of insulating materials due to the dielectric heating mechanism [1], • thereby an accelerated thermal aging [2], • a different partial discharge inception and extinction voltage [3], • an increased partial discharge intensity [4] and • an accelerated electrical field aging due to an superimposed alternating field [2]. The dielectric heating mechanism is one of the fundamental effects and relevant because the thermal energy of an insulation material is increased. Due to this, the possibility of insulation materials and insulation systems to withstand this combined thermal-electrical stress is of special interest.

Fig. 1: Real and synthesized mixed voltages for replication of voltage stress with the focus on dielectric heating

Jicable'15 - Versailles 21-25 June, 2015

1/5

B6.6

9th International Conference on Insulated Power Cables

MIXED VOLTAGE BREAKDOWN OF PVC

B6.6

transition temperature leads to the breakdown.

The breakdown voltage tests of PVC- (Polyvinylchloride) specimen are performed and shown in [6] with a test setup according to a standard electrode assembly in a homogeneous electrical field.

Fig. 3: FEM – Multiphysics simulation of thermal electrical stress during the breakdown test

DSC-ANALYSIS OF TEST-SPECIMEN

Fig. 2: Breakdown voltage of PVC specimen at mixed voltage with high frequency content (25 kHz) The measured breakdown voltages ukHz, BD compared to the breakdown voltage at 50 Hz (63,0 kV ± 2,2 kV [6]) show that the insulation strength is decreased due to the superimposed high frequency voltage. Figure 2 shows the measured breakdown voltages (ukHz, BD) in dependency of a fixed subordinated voltage (u50 Hz or uDC). If the subordinated voltage increases, the superimposed high frequency voltage has to be increased as well in order to initiate the breakdown. A FEM – Multiphysics simulation is used to calculate the theoretical temperature rise at the specimen caused by the heat source density due to dielectric heating (see Figure 3, left). This method is verified by temperature measurements in [7]. The velocity of temperature rise correlates with the voltage rise time of 6 kV/sec and increases with a characteristic function according to the heat transport mechanism (see Figure 3). The diagram shows that the glass transition temperature of PVC (70°- 80°C) isn’t reached at the breakdown voltage value of ukHz, BD = 11 kV (see Figure 2, left: u50Hz = 0 kV). This temperature is a maximum operation temperature for PVC, which should not be reached to guarantee the dielectric strength. The calculated and the measured values indicate that either the breakdown occurred below glass transition temperature or the temperature distribution was inhomogeneous and thereupon local overheating over the glass

The visual examination of the breakdown specimen has revealed that the surface exhibits several structural types after the breakdown test (see Figure 4). The breakdown always occurs at one point at structure A. Therefore the DSC – method (Differential Scanning calorimetry) is used to analyze different parts of the specimen with respect to the temperature reached. The aim is to find out if an inhomogeneous temperature distribution during the breakdown test occurs. This standard-method provides information of the thermal, mechanical and electrical prior history of a polymer. It is one of a fingerprinting method recommend in the CIGRÈ brochure 595 [8]. With it, the heat flow between a sample and a reference is measured. The specific heat capacity, enthalpies and characteristic temperatures such as glass transition and melting point can therefore be determined. Figure 5 and Figure 6 show the measured DSC-curves of two material samples A and B according to Figure 4 of the PVC specimen after the breakdown test with different voltage stress. The second heating curve refers to the chain dotted line. The glass transition temperature is determined at 77°- 78°C and the melting point at approximately 124°C. Both diagrams show no significant differences between the first and second heating curve of the material sample B. The first heating curve of sample B shows the typical trend which was also measured at electrical unstressed material samples. The first and second heating curve of the material sample A shows an enthalpy relaxation peak at different positions (grey shaded) in both diagrams. This is justifiable by the different heating and cooling velocities as well as different state of order in the microscopic structure of the polymer [9]. It indicates that in the volume of structure A (see Figure 4) the temperature was higher than the glass transition temperature and the cooling velocity after the breakdown test was high enough to freeze a non-energetic favorable state, i.e. local ordered structures due to the electrical field force.

Fig. 4: Enlarged view of the surface of a specimen with breakdown point and different structures after test

Jicable'15 - Versailles 21-25 June, 2015

2/5

B6.6

B6.6

9th International Conference on Insulated Power Cables

PROCESS BEFORE LOSS OF DIELECTRIC STRENGTH DUE TO DIELECTRIC HEATING To understand the process immediately before the loss of the dielectric strength of different polymers at mixed voltage stress, high-resolution voltage, current and temperature measurements are carried out with the electrode arrangement in Figure 7. Surface discharges are reliably prevented by encapsulating the electrode arrangement with SF6 – gas. The temperature ϑ is measured in the middle of the upper electrode. The thickness of the electrodes is one millimeter and the polymers are stressed below breakdown voltage. Due to the minimized heat transfer to the ambience, an imbalance thermal equilibrium is realized.

Fig. 5: DSC curves of PVC after breakdown due to stress with sinusoidal voltage (25 kHz) It should be noted, that after the breakdown occurs, first the high frequency voltage and secondly the subordinated voltage is switched off. A correlation between the integral of the enthalpy relaxation peak and the amplitude of the applied mixed voltage cannot be exactly determined. Furthermore the effectively acting electrical field strength and the maximum temperature of the sample are not determinable at the moment of passing the glass transition point downwards. The position of the enthalpy relaxation peak can be found to be reproducible at the same positions if a mixed voltage acts or not. Essentially the reason for the breakdown is local overheating of the stressed volume. The heating of the samples are induced by dielectric heating due to a superimposed high frequency voltage and the high loss factor of PVC.

Fig. 7: Electrode assembly For the analysis of the temperature dependent alteration of the electrical properties of the specimen during the test, a method according to [10] is used. The relative permittivity and resistance of the specimen, based on a RC-parallel electrical equivalent circuit, is calculated using complex calculation (see equation 1-5). In the equations shown the capacitance and resistance are calculated in each period of the measured signals. This is carried out for high frequency voltage stress and mixed voltage stress (direct voltage superimposed by high-frequency voltage). The relative permittivity can by calculated with the geometry of the specimen and the determined capacitance. The exact value of electrical conductivity cannot be calculated, due to non-negligible polarization losses. ∙ ∙

[2]

∙

[3] ! "

$

Fig. 6: DSC curves of PVC after breakdown due to stress with mixed voltages (25 kHz and 50 Hz)

[1]

# "

! "

# " ∙% ! "

# " &

[4] [5]

First the measured and calculated values of a POM(Polyoxymethylen) specimen are shown in Figure 8 over a measuring period of twenty seconds. The upper two diagrams show a zoom view of the measured high frequency voltage uHF and current iHF through the specimen. At the beginning of the measuring time the signals are sinusoidal, amplitudes are constant and the temperature of the specimen is 48°C. At the time of 2,5 s the voltage uHF decreases and is adjusted immediately. Until this time the calculated resistance R is lowered to 1 MΩ. Because of the unpolar character of POM (no temperature dependence of the relative permittivity), the relative permittivity εr has not changed. From now the amplitude of the current iHF increases and discharges occur (see also the zoom view in the upper right diagram)

Jicable'15 - Versailles 21-25 June, 2015

3/5

B6.6

9th International Conference on Insulated Power Cables

and the calculation of permittivity εr and resistance R is affected by these non-sinusoidal signals. The continuous increase in temperature at the end of the diagram is also reasonable with the heat input of the partial discharges. Only shortly after the measuring period, the discharges lead to the specimen’s breakdown.

B6.6

For POM and PMMA no difference can be found, if they are stressed with mixed voltage or not (direct voltage and superimposed high-frequency voltage). The thermal impact due to the dielectric heating is the basic cause for loss of dielectric strength. For PVC a different behavior is determinable, if a specimen is stressed with an additional direct electrical field or not (see Figure 10 and 11)

Fig. 9: Measured and calculated signals of a 0,5 mm PMMA specimen at stress with 10 kV at 16,5 kHz

Fig. 8: Measured and calculated signals of a 0,5 mm POM specimen at stress with 8 kV at 16,5 kHz However, three mechanisms which lead to the loss of dielectric strength are recognizable. First, the temperature increases due to the dielectric heating. Secondly the resistance of the sample decreases in relation to the increasing temperature and third, partial discharges occur between the electrodes and specimen which leads to the destruction of the material. The decreasing resistance can be correlated with an increasing electrical conductivity, which is highly sensitive to temperature change at polymers. Figure 9 shows the measured and calculated signals at PMMA (Polymethylmethacrylat), which has been determined by the same method over a measuring period of twenty seconds. By altering the temperature from 90°C to 110°C, a change of relative permittivity is identifiable. In this range the temperature exceeds the glass transition temperature of PMMA, which is linked with the initiation of the main-chains polarization of polymers (α-relaxation). This leads to a higher relative permittivity over the glass transition point which also correlates with the increasing current iHF through the specimen. The resistance R changes during the glass transition point with the shown curve reproducible several times, but cannot be interpreted clearly (time -6,5 s). After the measuring period, the breakdown occurs due to partial discharges which lead to the destruction of the material.

Fig. 10: Measured and calculated signals of a 0,5 mm PVC specimen at stress with 7,5 kV at 15 kHz Similar to the behavior of PMMA (see Figure 8), Figure 9 shows, an increasing relative permittivity εr and current iHF during the passing of the glass transition temperature between 70°C and 80°C (compare DSC-analysis in Figure 5 and Figure 6). It should be noted that there is little deviation between the inner specimen temperature and the temperature at the measuring point, which is caused by the heat transfer mechanism. Resistance R decreases due to the increasing specimen temperature. During the glass transition a further decrease of the resistance R is noticeable.

Jicable'15 - Versailles 21-25 June, 2015

4/5

B6.6

9th International Conference on Insulated Power Cables

B6.6

Figure 11 shows the measured and calculated signals over the same temperature range, but twice measuring time than in Figure 10. The applied direct voltage uDC is also shown. Due to the decreasing resistance, the direct voltage uHF decreases during the glass transition temperature, because it is not adjusted.

point. For PVC different behavior was found, but nevertheless the failure of the specimen occurs. Basically the electrical conductivity increases rapidly with increasing temperature. Further, partial discharges occur due to different thermal expansion of insulation materials or adjacent electrodes or materials.

It is remarkable that during the glass transition temperature the relative permittivity εr has not so highly increased as in Figure 10, where the specimen is not stressed with a direct electrical field. The current iHF also moderately increases. This leads to the assumption that the relaxation mechanism of PVC over the glass transition point is influenced by an additional direct field.

To understand the electrical strength at mixed voltages it is helpful to comprehend the resistive-capacitive behavior of the used insulation materials. At polar materials, such as PVC or PMMA, the dielectric losses have to be considered more than at non-polar materials. However, the electrical conductivity is a temperature sensitive value and low heating can lead to increased electrical conductivity. Furthermore the electrical field distribution is affected by the changing of electrical material properties.

After the measuring time the specimen are also affected by partial discharges, which leads to its breakdown.

REFERENCES

Fig. 11: Measured and calculated signals of a 0,5 mm PVC specimen at mixed voltage stress with 7,5 kV at 15 kHz and a direct voltage of 8 kV

CONCLUSION Due to the increasing application of converter technologies in our entire energy grid, electrical insulation systems have to withstand different mixed voltage forms with high-frequency content. To understand the impact of superimposed dielectric and conductive fields, as well as different amplitudes and frequencies, basic investigations and the testing of electrical equipment should be done with well-defined mixed voltages. The form of mixed voltages as a test voltage depends on the effect which is the focus of interest. At different polymers the measured and calculated signals under thermal-electrical stress have shown that highfrequency components of the used mixed voltages are the primary reason for the loss of dielectric strength. At polymers where the operation temperature is below the glass transition temperature, intense heating leads to a change of material properties during the glass transition

[1] M. Birle, C. Leu, 2011, “Dielectric heating in insulating materials at high DC and AC voltages superimposed by high frequency high voltages”, XVIII International Symposium on High Voltage Engineering (ISH), Seoul, 2011 [2] R. Patsch, J. Kindersberger, D. König, 2002, “Alterung von Betriebsmitteln - ein Überblick”, ETG Fachbericht 87 Diagnostik elektrischer Betriebsmittel, Berlin, 2002 [3] E. Lindell, T. Bengtsson, J. Blennow, S.M. Gubanski, 2010, “Influence of Rise Time on Partial Discharge Extinction Voltage at Semi-square Voltage Waveforms” IEEE Transactions on Dielectrics and Electrical Insulation Vol. 17, No. 1; February 2010 [4] M. Paede, W. Pfeiffer, R. Plessow, “Testing of insulation systems with respect to high-frequency voltage stress” European Transaction on Electrical Power (ETEP), Vol. 12, No. 5, pp. 337–345, September-October, 2002 [5] M. Birle, C. Leu, S. Bauer, “Design and Application of a high-frequency high-voltage generator” XVII International Symposium on High Voltage Engineering (ISH), Hannover, 2011 [6] M. Birle, C. Leu, 2013, “Breakdown of polymer dielectrics at high direct and alternating voltages superimposed by high frequency high-voltages”, IEEE International Conference on Solid Dielectrics (ICSD), 2013 [7] E. Wettengel, 2014, “Gekoppelte Simulation von elektrischem und thermischem Feld eines Isoliersystems mit FEM“, Master Thesis, Technische Universität Ilmenau, Germany, 2013, [8] Cigré WG D1.27, 2014, “Cigré Brochure 595: Fingerprinting of polymeric insulating materials for outdoor use”, Oktober 2014 [9] W. Knappe., H.J. Ott, 1977, “Die spezifische Wärme von PMMA und PVC im Einfrierbereich“ Colloid & Polymer Science. 255, 837-843, 1977 [10] C. Leu, M. Birle, S. Gossel, F. Reichert, 2014, “Determination of electric conductivity of hot argon gas using a high-frequency high-voltage generator”, XXth International Conference on Gas Discharges and their Applications, Orléans, 2014

Jicable'15 - Versailles 21-25 June, 2015

5/5