Comparison of Mechanical and Low-Frequency Dielectric Properties of Thermally and Thermo-Mechanically Aged Low Voltage CSPE/XLPE Nuclear Power Plant Cables

: During the service period of low-voltage nuclear cables, multiple stresses inﬂuence the aging of polymeric materials of cables. Thermal and radiation stresses are considered service aging factors in qualiﬁcation tests, while the standards usually do not prescribe mechanical stress. CSPE/XLPE insulated nuclear cable samples were exposed to thermal and combined thermo-mechanical aging for more than 1200 h at 120 ◦ C. The real and imaginary parts of permittivity were measured in the 200 µ Hz to 50 mHz range as dielectric properties. The Shore D hardness of the samples was measured to analyze the mechanical characteristics of the cable. To characterize the dielectric spectrum, derived quantities, namely central real and imaginary permittivities and real and imaginary permittivities’ central frequencies were calculated. The change of dielectric spectra did not show a clear trend with aging, but the imaginary permittivity’s central frequency was higher by 0.5 mHz in the case of thermo-mechanically aged samples. The Shore D hardness was also higher on the thermo-mechanically aged samples. These ﬁndings show the combined aging has a higher impact on the insulation properties. Hence, involving the mechanical stress in the aging procedure of cable qualiﬁcation enables the design of more robust cables in a harsh environment.


Introduction
In 2020, the built-in nuclear-generating capacity was 393 GW (electrical), and its contribution to electricity generation is expected to increase. In the highest case predictions, the world's nuclear capacity will be doubled to 792 GW by 2050, keeping its almost 5% ratio [1]. Hence, nuclear power will keep its important role in power generation for decades, and the current nuclear power plants' (NPPs') designed lifetime is potentially extended to 80 years [2]. During this long period, the safe and reliable operation of NPPs in normal service and design basis events (DBEs) require safety-related equipment performance to meet their designed requirements. The equipment qualification is the procedure when safety-related equipment performance is tested in all operational states and accident conditions [3]. Since the cabling condition plays a key role in the reliable operation of safety systems, the safety-related cables as system components are also subject to the qualification which the test procedure is described by standards [4]. More the 1000 miles of cable installed typically in a pressurized water reactor (PWR), low-voltage (LV) power cables account for only 15% of the total LV cables deployed in the NPPs [5]. Still, they are highly important as they supply motors and pumps related to nuclear safety. The structure of these

Specimens
A Firewall III radiation-resistant Class 1E low-voltage (600 V) NPP unshielded power cable (RSCC Wire and Cable, East Granby, CT, USA) under investigation is shown in Figure 1. The cable consisted of three parts: annealed, tin-coated copper conductor with a cross-sectional area of 6 AWG (13.3 mm 2 ), cross-linked polyethylene (XLPE) core insulation having a thickness of 45 mils (1.143 mm), and chlorosulphonated polyethylene (CSPE) jacket with 30 mils (0.762 mm) thickness. The overall diameter of the cable was 0.34 inches (8.636 mm). Under the International Atomic Energy Agency's (IAEA's) guidelines [35] and as expressed in Figure 1, 1 cm and 3 cm of the insulation and jacket were removed from both sides of the cable sample, respectively. The cable sample was 0.5 m in length for the experiments.

Thermal Aging
The cable samples were exposed to 120 • C accelerated thermal agings in an aircirculating oven for seven different periods: 176, 338, 512, 784, 912, 1056, and 1248 h. Applying these aging parameters is because the 120 • C aging temperature was also used for benchmark analysis of this cable type in the IAEA Coordinated Research Project [35]. Moreover, elongation at break data for the CSPE jacket are also available for these aging times in [35]. The cable samples in the oven are shown in Figure 2a.

98
The cable samples were exposed to 120°C accelerated thermal aging in an air circulated oven for 99 seven different periods: 176, 338, 512, 784, 912, 1056, and 1248 hours. The 120°C aging temperature 100 was also used for benchmark analysis of this cable type in a IAEA Coordinated Research Project [16].

101
The cable samples in the oven are shown in Figure 2 (a).

Thermo-Mechanical Aging
According to the IEEE standard, the cable samples were bent on a mandrel with an outer diameter of 15 cm for the thermo-mechanical (T-M) simultaneous aging [4].
Then the cable samples were placed in the air circulated oven at a temperature of 120 • C. The cables were exposed to simultaneous aging for 176, 338, 507, 779, 907, 1051, and 1243 h (Figure 2b). After each aging period, the samples were removed from the cylinder and straightened for measurements. After the measurements, the samples were bent on the cylinder again. Then, they inserted back to the aging chamber for the next round.

Dielectric Spectroscopy Measurement
Since the subject of the investigation is a single core unshielded power cable having a jacket and core insulation also, this cable structure can be considered a layered insulation arrangement. The most dominant polarization process in layered insulations is interfacial polarization, with a far lower characteristic frequency than 1 Hz. Therefore dielectric spectroscopy was carried out in the frequency range of 200 µHz to 50 mHz by investigating the complex permittivity of the material. The complex permittivity is defined by the expression: where ε and ε are defined as the real and imaginary parts of permittivity since the former depicts the strength of the dipoles in the material aligning themselves when an external electric field is applied. This part is known as permittivity characterizing the electric energy stored within the dielectric material [36,37]. The imaginary permittivity represents the losses in the material due to conduction and polarization. The dissipation factor or tan δ is the ratio of the imaginary and real parts of permittivity (tan δ = ε /ε ). Modeling a relaxation process by the Debye model, the real and imaginary parts of permittivity as a function of angular frequency can be expressed by the following equations: where the ε s and ε ∞ are the permittivities at 0 and infinity frequency, at the same time, ω 0 is the characteristic angular frequency of the relaxation process. The relationship between ε s and ε ∞ can be seen in Figure 3. As the figure shows, the imaginary part of permittivity has a peak at the characteristic frequency, i.e., loss peak. Where the imaginary part has the peak, the changing of the real part is the highest. By changing the characteristic frequency, the curves shift together. The complex permittivity also enables the determination of the conductivity of the material. However, ε contains the sum of resistive and dielectric losses, if the resistive part is dominant, the slope of ε(ω) is ω −1 and ε (ω) is constant in the dielectric spectrum [38]. At very low frequencies (below 1 Hz), the dc conduction is the dominant part of dielectric loss in XLPE cable insulation [39].
The measurements were carried out with Omicron Dirana (Omicron Electronics Gmbh, Klaus, Austria). This equipment uses the combination of frequency-domain spectroscopy (FDS) and polarization-depolarization current (PDC) methods to determine the dielectric response. The equipment directly measures the dielectric response in the frequency domain at higher frequencies by the FDS technique. To determine the response in the lower frequency range, time-domain PDC data are recorded. Then, the PDC data are converted to the frequency domain by using discrete Fourier transformation. This technique has an important advantage, namely, it reduces the measurement time by 73% [38,40]. The accuracy of the equipment is 2% below 1 mHz for power and dissipation factor measurement and 0.5% for capacitance.   Since the subject of the investigation is a single core unshielded cable, the measurement of dielectric properties requires an external electrode, which was prepared by wrapping alumina foil around the cable. The voltage source, the equipment terminal, was connected to the external foil electrode through a copper strip. In contrast, the current output was connected to the internal conductor, as the measurement setup is shown in Figure 4. This connection ensures the lowest noise in the result. The cable samples were placed in the Faraday's cage connected to the guard for more noise suppression.

161
After each measurement, the cable samples were discharged for 5 hours before the start of the next 162 aging period.

171
The hardness of the cable was measured at the ends and center with a total of 10 measuring

Shore D Hardness Measurement
The hardness measurement is based on the indentation depth of the indenter foot into the material applying a constant force. If the indentation is the highest, the measured value is 100, and if there is no indentation, the result is 0. In the case of the Shore D scale, the force is 4450.0 mN, the maximum penetration distance is 2.54 mm, and a 30-degree conical pressure foot is used. The Shore D hardness measurement is suitable for testing flat samples with a minimum 6 mm thickness according to the ASTMD2240-05 standard [41]. Therefore in this research, the Shore D hardness was tested as a comparative measurement to investigate the mechanical properties.
The hardness of the cable was measured at the ends and center with a total of 10 measuring points at 25 ± 0.5 • C. Although the test method is suitable for the flat samples [41], with the adaption of the foot adapter, the measurement can be executed for the round surfaces. In this investigation, the Shore D hardness measures the resultant hardness of the jacket and the insulation.

Thermal Aging
The measurement result of the real part of permittivity (ε ) for the thermal stress is plotted against the frequency for each period in Figure 5a. According to the general behavior of dielectrics, the ε decrease as the frequency increase. This characteristic behavior of the ε was the same for all the aging periods. With aging time, the ε either decreased or remained constant at all frequencies, except at 2 × 10 −2 Hz and above, where a slight increase was observed. This behavior can be explained by the comparison with the imaginary parts curves. The results of the imaginary part of permittivity (ε ) are plotted in Figure 5b. The rearrangement of curves in the frequency range suggests that the ε curves shift towards a lower frequency range with aging, indicating the conductivity decrease with aging. This behavior explains the decreasing of ε values at the lowest frequencies. In the case of the unaged sample, the ε started with a certain value, then increased with the increase in frequency and reached a maximum value, and then again started to decrease. With thermal aging, the ε increased between 2 × 10 −4 Hz and 1 × 10 −3 Hz for the first two thermal aging periods, and then ε decreased with aging time. For frequency higher than 1 × 10 −3 Hz there was a decreasing trend in the values of ε . This behavior was observed until the sixth thermal aging period (1056 h), whereas the values increased at all frequency points.

Thermo-Mechanical Aging
The ε in the case of the T-M aging is shown in Figure 6a. The behavior of the ε against frequency was similar to the thermal stress, irrespective of the aging. In contrast, the observed impact of T-M was as the ε increased after the first two periods for the whole frequency spectrum, while it started to decrease after the third period (507 h). A slight increase after the sixth period (1051 h) was observed, while a substantial decrease in the whole curve was noted after the seventh T-M period. The frequency plot of ε values is shown in Figure 6b. For almost all aging periods, ε started with a particular value. Then, the imaginary part of permittivity increased with frequency, reached a peak value at a specific frequency, and then declined with a further frequency increase. A noticeable increase in the values was noted in the ε after the first T-M period between 2 × 10 −4 Hz and 1 × 10 −3 Hz, while at higher frequencies, the values decreased. After a longer aging time, the ε decreased at all frequencies till the fifth T-M period (907 h). Like the ε , after 1051 h, the ε increased, which then decreased after the last aging round.

Shore D Hardness
The Shore D hardness recorded results of the cable under thermal and thermomechanical aging have been plotted in Figure 7. An overall increase in the hardness of the cable was also observed for thermal and thermo-mechanical aging. However, the hardness increase is higher in thermo-mechanical aging, which suggests the combined stress has a more intensive effect on the material structure, which results in higher hardness values.

Discussion
The increase in ε in a very-low-frequency range (below 1 Hz) is associated with the electrode polarization when the charge carriers accumulate near the electrodes, increasing the apparent value of the dielectric constant [42]. While the ε is a component related to the dielectric loss [43,44], one can expect that with aging, an increase in the losses will be experienced with the degradation of the polymer. Three frequencies were selected to simplify the analysis of the results: 200 µHz, 1 mHz, and 10 mHz, and the behavior of both kinds of aging were studied. The plot of the ε and ε for each stress at the selected frequencies is shown in Figure 8. Under thermal aging, the ε at the selected frequencies decreased with aging. After the last aging period, it increased, at 10 mHz, the ε value even exceeded the unaged value.
Under T-M aging, at 200 µHz and 1 mHz, an initial increase in the ε was observed, then the ε started to decrease with more aging times. In contrast, different behavior is seen at 10 mHz, where the real part of permittivity dropped after the first period. Then ε increased until the third period. More aging periods resulted in a decrease in the ε values. A surprising behavior was observed under the T-M aging after the sixth period, where the values of the ε increased even higher than the unaged values, which then decreased after the seventh period.
The effect of thermal stress on the ε caused an increase in the values for the first two periods at only 200 µHz and 1 mHz, which then declined with more aging, but at 10 mHz the values linearly decreased. The last thermal aging period resulted in a substantial increase in the values. At 200 µHz the value of ε remained higher than the unaged ones for all the aging periods.
While the T-M stress resulted in the increase in the ε after the first stress period at 1 mHz and 10 mHz, then it decreased with aging. In comparison, a clear trend was not observed at 200 µHz, where only the first and third periods affected the increase in the values higher than the unaged ones. Like the ε , the ε values increased at all the selected frequencies after the sixth thermal period, whereas after the seventh thermal period, the values decreased.
The loss peak frequencies are also plotted for thermal and thermo-mechanical aging ( Figure 9). For an unaged cable sample, a polarization peak was observed at 1 mHz, which with aging, was shifted to 200 µHz Figure 9a. Nevertheless, the curves of Figure 5b suggest in the case of aged samples, the peak of the polarization peak should be below 200 µHz. In the case of the T-M stress, the polarization peak did not change its position after the first aging period. However, it shifted to a higher frequency of 2 mHz until the fourth aging period and then to 200 µHz after the fifth aging period, Figure 9b. At the same time, it shifted back to 1 mHz after the sixth period and remained as it for the seventh period.

251
where only the first and third periods affected the increase in the values higher than the unaged ones.

252
Like the ', after the sixth thermal period, the values increased at all the selected frequencies,

253
whereas after the seventh thermal period the values decreased.

254
For the case of an unaged cable sample, a polarization peak was observed at 1x10 -3 Hz which 255 with aging, was shifted to 200x10 -6 Hz for the , Figure 9       For a more detailed analysis of the dependence of real and imaginary parts of permittivity curves on aging, and similarly to [30,45], four deducted quantities have been calculated, namely the central real and imaginary parts of permittivity (CRP and CIP), and the real and imaginary permittivities' central frequencies (RPCF and IPCF). The CRP is calculated by summing up the multiplication of the logarithm of the frequencies by the measured ε values at given frequencies and dividing this sum by the sum of the logarithm of the frequencies, as it can be seen in Equation (4).
By changing the ε to ε in Equation (4), the CIP can be calculated: The RPCF can be calculated as the average of the permittivity values weighted by the logarithm of the frequency: By changing the ε to ε in the Equation (6), the IPCF can be calculated: These deducted quantities make spectroscopy curves evaluate easier since two numbers characterize the shifting of the curves. CRP and CIP are the characteristic values for real and imaginary permittivity changes over the frequency range of investigation. RPCF and IPCF characterise the shift of the curves in the frequency range. The change of CRP RPCF with aging is Figure 10.  Figure 10a shows that CRP of thermally aged samples decreases with aging from the initial 3.2 below 3. However, after the last aging round, it jumps up to almost 3.15. In the case of thermo-mechanically aged samples, the CRP has increased over 3.3 after the first aging round. Then, it decreases. The fitted trendlines show that the real part of the permittivity for the T-M samples was higher by nearly 0.1 during the whole aging period in the investigated frequency range.
The CRP curve of thermally aged samples shows a monotonous decreasing trend with aging, while the CRP of thermo-mechanically aged samples does not change (Figure 10b). Accordingly, the curves of the real part of permittivity of thermally aged samples shift towards lower frequencies, while there is no significant change in the case of T-M samples.
In general, the CIP and IPCF values clearly show the changing of an average loss in the investigated frequency range and the changing of the average polarization peak with aging. The values of CIP and IPCF are depicted in Figure 11. Figure 11a shows the CIP change with aging in case thermal and T-M aging. In both cases, the dielectric loss increases at the beginning of aging. The CIP increases in the first two aging periods of thermal aging. A more significant increase can be observed in the first aging period of T-M aging, then the CIP values decrease. By calculating the linear trend, no significant difference can be observed between thermal and T-M CIP values.
The IPCF values (Figure 11b) were dropped by 0.5 mHz after the first aging round, then the IPCF of thermally aged samples decreased, while that of T-M samples shows a small variation, a minimal decreasing trend can be observed. The calculated linear trends suggest the IPCF values of T-M samples are higher 0.5 mHz at the ending of aging. The trendlines show the ε curves shifted towards higher frequencies in the case of T-M aging. Hence the polarisation peaks are at higher frequencies than the thermally aged samples. Under the thermal stress in the presence of oxygen, it is believed that the polymer undergoes molecular structural changes due to the cross-linking, chain scission, and oxidation reactions, which occur at the same time in the crystalline and amorphous regions. The actual cause of the degradation of the polymer depends on the dominance of one of these reactions. The cross-linking reaction generates a solid network that opposes the external field's effect and restrains the dipoles' movements, which is a case opposite to the chain scission reaction.
However, a very simple cable was chosen for the experiment, from the point of view of aging, the subject of the experiment is a complex structure. It contains two layers made from different polymers; moreover, the inner XLPE is covered by the CSPE jacket. Therefore, during the accelerated aging, the XLPE insulation was not contacted directly with the air. By investigating cable jacket samples removed from commercial cables, the significant effect of diffusion-limited oxidation (DLO) on the degradation of CSPE cable jacket material has been reported at temperatures as low as 110 • C [46]. Consequently, in the case of intact cable samples, when the air contacts only the outer surface of samples, it can be assumed the CSPE jacket consumes oxygen during aging at elevated temperatures. Therefore, the oxidative degradation of the XLPE insulation was very limited; hence, cross-linking and recrystallization were the main degradation mechanisms [47].
In addition to these reactions, materials such as CSPE, which has chlorine and sulfur dioxide attached to the polyethylene backbone, undergo dehydrohalogenation. During dehydrohalogenation, the halogen atom is separated from the polymer chain and the hydrogen, resulting in double bonds in the polymer chain [48,49], resulting in increased conductivity and permittivity [50].
Under the exclusive mechanical stress, it has been reported that the polymer may experiences two stresses: tension and compression [14]. The former assists the movement of the molecular chain, which may lead to the fracture of the chain and could increase the microcavities size [51,52]. It has also been reported that this would also increase the space charges. In contrast, the compression stress acts opposite to the tension, where the molecular chains come close to each other and support the bond attraction leading to more stabilization. This also inhibits the increase in the intensity of space charges. Since in our case, the cable has been exposed to both thermal and mechanical stresses, where the inner side of the cable is under compression force and the outer side under tension force, Figure 12. Therefore, cross-linking and chain scission reactions in the presence of oxygen under thermal stress are also playing their role. The cable insulation's hardness increase under both types of stresses is also related to the reason that at elevated temperature, the generated small molecules due to decomposition reactions re-cross link with each other and increase the cable's hardness [47]. This phenomenon in the presence of compression stress due to mechanical aging is reinforced.
From the point of dielectric measurement view, the insulation and the jacket of nonshielded, single-core cables are considered a layered arrangement. Hence, the applied electrode arrangement (covering the cable's surface by conductive foil) can investigate only the resultant dielectric properties, i.e., the real and imaginary parts of permittivities, of the layers. Similarly, the Shore D hardness also measures the resultant hardness of both layers. Since the degradation processes also increase the hardness of the CSPE and the XLPE, the result will also be an increasing trend in thermal and T-M aging cases. However, conductivity and permittivity of CSPE increase due to thermal aging, the resultant dielectric loss and permittivity decrease in the investigated frequency range according to the change of CIP values. Nevertheless, degradation processes of the XLPE, namely the cross-linking and recrystallization, decrease its conductivity and permittivity [53]. Since the thickness of the XLPE insulation is higher by 50%, the resultant values of the real and imaginary parts of permittivities are significantly affected by the XLPE. The annealing effect of XLPE can cause the initial increase in CIP. The annealing effect increases the conductivity of XLPE at the beginning of thermal treatment [54].
It can be concluded that the parallel thermal-mechanical stress results in more intensive degradation of the mechanical properties of cable insulation than the thermal stress alone. Since the layered structure of jacket and insulation makes it difficult to identify the degradation processes and analyze the dielectric behavior. Further studies, including chemical analysis, are needed to reveal the degradation mechanism of single-core LV cables. This study would be beneficial to point out the importance of applying combined mechanical-thermal stressing for accelerated aging testing in the qualification procedure of NPP cables.

Conclusions
The main purpose of the work was to compare the dielectric behavior in-low-frequency range and Shore D hardness of thermally and thermo-mechanically aged single-core CSPE/XLPE insulated LV NPP cables. For the purpose of investigation, one group of samples was straightened, and the other one was coiled on a cylinder. Both groups of samples were aged at 120 • C for more than 1200 h. The dielectric spectrum in the 200 µHz to 50 mHz range and the Shore D hardness were measured on both groups. By analyzing the dielectric spectra, the real and imaginary parts of permittivities changed differently in both groups although, they did not show a clear trend with aging. To clarify the changes in the dielectric spectrum due to aging, derived quantities, namely, the central real and imaginary parts of permittivity and the real and imaginary permittivities' central frequency were introduced. The deducted quantities show that the real part of the permittivity of the thermo-mechanically aged samples was higher by nearly 0.1 than that of thermally aged ones during the whole aging period in the investigated frequency range. The deducted quantities show that the real part of the permittivity of the thermo-mechanically aged samples was higher by nearly 0.1 than that of thermally aged ones during the whole aging period in the investigated frequency range. At the same time, the central frequency of the real part of permittivity curves decreased by 1 mHz due to thermal aging. A similar change was not observed on thermo-mechanically aged samples.
However, the central imaginary part of permittivity did not differ significantly between the groups, the imaginary permittivity's central frequency was higher by 0.5 mHz in the case of thermo-mechanically aged samples. This suggests a higher conductivity of thermo-mechanically aged samples, which can be the result of the annealing effect. During aging, the Shore D hardness was also higher on the thermo-mechanically aged samples than the thermally aged ones. At the end of aging, the hardness of the thermo-mechanically aged samples was 34% higher than the initial value, while that of the thermally aged samples was only 30% higher.
Based on the results, it can be concluded that mechanical stress has a strong impact on the degradation of the polymeric materials of the cable in combination with the thermal stress, which is considered during the simulation of service aging of cables. These results could lead to the inclusion of mechanical stress in the aging procedure of cable qualification, enabling the design of more robust cables in a harsh environment. For the future perspective, it is intended to carry out the chemical and other electrical and mechanical tests to further elaborate and understand the impact of mechanical stress with thermal stress on the LV cable aging.

Data Availability Statement:
The data presented in this study are available on request from the corresponding author.

Conflicts of Interest:
The authors declare no conflict of interest.