In the paper, the performances of XLPE and EPR as materials candidate for insulation of high voltage power cables are investigated, via the analysis of a large amount of short and long-term life tests. The tests were realized applying electrical, thermal and thermoelectrical stresses on cable models of reduced size. The use of cable models makes test procedures simpler and cheeper, while retaining the geometry of power cables. This enables an easier extrapolation to the service conditions of power cables, since only the size effect has to be accounted for. All test cells and procedures are described in the paper, including the so-called progressive censoring procedure, applied to electrical and multistress life tests to obtain information on the aging mechanisms correlated to failure data, as well as to reduce test times. The life tests were performed with the purpose to enable the determination of the endurance characteristics of the materials in a stress range as close as possible to the service one, according to a strategy that allows optimization between the need of reasonably-short test times and characterization of the low stress behavior. The analysis of the results has the purpose to compare the thermo-electrical endurance characteristics of the two materials, and point out the possible advantages that each one has in the perspective of the use as insulation for HV cables. In order to reach this goal, the experimental data are processed by means of appropriate statistical tools. Since life data are relevant to given failure probabilities, appropriate failure probability distributions (e.g. the Weibull one) are used to assess life percentiles. Then, single and multistress life models (able to fit the life points on the relevant graphs) allow the derivation of endurance indices and the subsequent extrapolation to service stress levels, together with the drawing of life lines and isochronal lines. These lines provide a synthetic and exhaustive description of the behavior of the tested materials. The Combined Analysis Method (CAM) aids in drawing life and isochronal lines, and also enables the choice of the optimum electrical, thermal, and electro-thermal test stress levels. The thermal life tests were performed according to IEC 216 standard. Different test temperatures, diagnostic properties (e.g. electric and tensile strength, weight, density) and end-points were considered, in order to achieve a thermal endurance evaluation homogeneous to the results of multistress tests at low electrical field. The Arrhenius model fits well the experimental points for both XLPE and EPR cable models. However, the test results suggest the existence of a thermal threshold, with a value of about 100 and 105 degrees C for XLPE and EPR, respectively. The values of the thermal endurance indices, TI = Temperature Index (HIC = Halving Interval in Celsius) have been derived, respectively, for XLPE and EPR: 101(7.8) and 111(8.4). From these results, it derives that the tested EPR shows, on the whole, better thermal endurance characteristics with respect to the tested XLPE, as shown by the higher TI value and the higher guessed thermal threshold. The electrical life tests were carried out at 50 Hz (plus a few high-frequency tests to investigate the long-term trend of the electrical life lines). Even if EPR shows a slightly lower initial electric strength, it has considerably higher Voltage Endurance Coefficient (VEC), thus apparently better endurance characteristics. However, in the case of XLPE cables an electrical threshold of approximately 11.5 kV/mm was detected at 20 degrees C (two threshold models, the 4-parameter Exponential Threshold Model, 4p-ETM, and the 4-parameter Inverse Power Threshold Model, 4p-IPTM, fit the electrical data), while in the case of EPR cables a threshold was not detected (both the exponential model, EM, and the inverse power model, IPM, fit the electrical life data). Hence, the linear behavior of EPR would require probabilistic life models in order to infer design stresses at the chosen probability, and, moreover, to rescale the test results to the power cable size. On the contrary, in the case of XLPE a value of the design electrical stress lower than the laboratory-detected threshold would ensure that the failure probability is practically zero, independently of failure probability and cable size. The thermo-electrical life test results relevant to XLPE cables fit well a parametric form of the 4p-ETM (temperature and probability are added parameters), from which the 4-parameter Exponential Threshold Probabilistic Model derives. A temperature-dependent threshold is detected, which disappears over about 100 degrees C. The choice of an electric stress value lower than the electrical threshold at the service temperature would free the design from probabilistic considerations. Multistress life models are quite complex but have the advantage of predicting life at temperatures different from the test ones. In the paper, an expression of the electro-thermal threshold life model is recalled, which takes advantage of accurate expressions of the temperature dependence of the parameters (in particular of the threshold) and fits quite well the experimental results relevant to XLPE cables. The isochronal lines for long lives provide a graphical index useful for material characterization, the Stress Compatibility Index, SCI. In the case of the tested XLPE cables, a value of SCI = 1.3 is obtained. The thermo-electrical life test results relevant to EPR show a linearity range (30-60 kV/mm, 20-100 degrees C) where the IPM (in a parametric form with respect to temperature and probability) well fits the experimental results. The VEC is still high even for temperatures exceeding the design one, thus indicating good endurance of the tested EPR to combined electro-thermal stress. The experimental results support the use and the validity of a phenomenological linear combinedstress life model, which can also be derived on the basis of thermodynamic considerations. The CAM points in the thermal and electrical life graphs (and the isochronal lines at long times) suggest the existence of an electrical threshold which could range from about 20 kV/mm at 90 degrees C, to 30 kV/mm at 20 degrees C. The very good multistress endurance characteristics of the tested EPR is proved by the isochronal graph, from which a value of SCI ~ 1.6 is obtained. A final comparison made on the basis of the endurance indices shows, on the whole, slightly better behavior of EPR with respect to XLPE, which is confirmed by a specific application to power cables. However, this result is valid for the tested materials and must be carefully extended to the two materials in general, particularly in the case of EPR.
Mazzanti G., Montanari G.C. (1997). A comparison between XLPE and EPR as insulating materials for high voltage cables. IEEE POWER ENGINEERING REVIEW, 17(1), 36-36 [10.1109/MPER.1997.560656].
A comparison between XLPE and EPR as insulating materials for high voltage cables
Mazzanti G.;Montanari G. C.
1997
Abstract
In the paper, the performances of XLPE and EPR as materials candidate for insulation of high voltage power cables are investigated, via the analysis of a large amount of short and long-term life tests. The tests were realized applying electrical, thermal and thermoelectrical stresses on cable models of reduced size. The use of cable models makes test procedures simpler and cheeper, while retaining the geometry of power cables. This enables an easier extrapolation to the service conditions of power cables, since only the size effect has to be accounted for. All test cells and procedures are described in the paper, including the so-called progressive censoring procedure, applied to electrical and multistress life tests to obtain information on the aging mechanisms correlated to failure data, as well as to reduce test times. The life tests were performed with the purpose to enable the determination of the endurance characteristics of the materials in a stress range as close as possible to the service one, according to a strategy that allows optimization between the need of reasonably-short test times and characterization of the low stress behavior. The analysis of the results has the purpose to compare the thermo-electrical endurance characteristics of the two materials, and point out the possible advantages that each one has in the perspective of the use as insulation for HV cables. In order to reach this goal, the experimental data are processed by means of appropriate statistical tools. Since life data are relevant to given failure probabilities, appropriate failure probability distributions (e.g. the Weibull one) are used to assess life percentiles. Then, single and multistress life models (able to fit the life points on the relevant graphs) allow the derivation of endurance indices and the subsequent extrapolation to service stress levels, together with the drawing of life lines and isochronal lines. These lines provide a synthetic and exhaustive description of the behavior of the tested materials. The Combined Analysis Method (CAM) aids in drawing life and isochronal lines, and also enables the choice of the optimum electrical, thermal, and electro-thermal test stress levels. The thermal life tests were performed according to IEC 216 standard. Different test temperatures, diagnostic properties (e.g. electric and tensile strength, weight, density) and end-points were considered, in order to achieve a thermal endurance evaluation homogeneous to the results of multistress tests at low electrical field. The Arrhenius model fits well the experimental points for both XLPE and EPR cable models. However, the test results suggest the existence of a thermal threshold, with a value of about 100 and 105 degrees C for XLPE and EPR, respectively. The values of the thermal endurance indices, TI = Temperature Index (HIC = Halving Interval in Celsius) have been derived, respectively, for XLPE and EPR: 101(7.8) and 111(8.4). From these results, it derives that the tested EPR shows, on the whole, better thermal endurance characteristics with respect to the tested XLPE, as shown by the higher TI value and the higher guessed thermal threshold. The electrical life tests were carried out at 50 Hz (plus a few high-frequency tests to investigate the long-term trend of the electrical life lines). Even if EPR shows a slightly lower initial electric strength, it has considerably higher Voltage Endurance Coefficient (VEC), thus apparently better endurance characteristics. However, in the case of XLPE cables an electrical threshold of approximately 11.5 kV/mm was detected at 20 degrees C (two threshold models, the 4-parameter Exponential Threshold Model, 4p-ETM, and the 4-parameter Inverse Power Threshold Model, 4p-IPTM, fit the electrical data), while in the case of EPR cables a threshold was not detected (both the exponential model, EM, and the inverse power model, IPM, fit the electrical life data). Hence, the linear behavior of EPR would require probabilistic life models in order to infer design stresses at the chosen probability, and, moreover, to rescale the test results to the power cable size. On the contrary, in the case of XLPE a value of the design electrical stress lower than the laboratory-detected threshold would ensure that the failure probability is practically zero, independently of failure probability and cable size. The thermo-electrical life test results relevant to XLPE cables fit well a parametric form of the 4p-ETM (temperature and probability are added parameters), from which the 4-parameter Exponential Threshold Probabilistic Model derives. A temperature-dependent threshold is detected, which disappears over about 100 degrees C. The choice of an electric stress value lower than the electrical threshold at the service temperature would free the design from probabilistic considerations. Multistress life models are quite complex but have the advantage of predicting life at temperatures different from the test ones. In the paper, an expression of the electro-thermal threshold life model is recalled, which takes advantage of accurate expressions of the temperature dependence of the parameters (in particular of the threshold) and fits quite well the experimental results relevant to XLPE cables. The isochronal lines for long lives provide a graphical index useful for material characterization, the Stress Compatibility Index, SCI. In the case of the tested XLPE cables, a value of SCI = 1.3 is obtained. The thermo-electrical life test results relevant to EPR show a linearity range (30-60 kV/mm, 20-100 degrees C) where the IPM (in a parametric form with respect to temperature and probability) well fits the experimental results. The VEC is still high even for temperatures exceeding the design one, thus indicating good endurance of the tested EPR to combined electro-thermal stress. The experimental results support the use and the validity of a phenomenological linear combinedstress life model, which can also be derived on the basis of thermodynamic considerations. The CAM points in the thermal and electrical life graphs (and the isochronal lines at long times) suggest the existence of an electrical threshold which could range from about 20 kV/mm at 90 degrees C, to 30 kV/mm at 20 degrees C. The very good multistress endurance characteristics of the tested EPR is proved by the isochronal graph, from which a value of SCI ~ 1.6 is obtained. A final comparison made on the basis of the endurance indices shows, on the whole, slightly better behavior of EPR with respect to XLPE, which is confirmed by a specific application to power cables. However, this result is valid for the tested materials and must be carefully extended to the two materials in general, particularly in the case of EPR.I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.
