1. Introduction
Power transformers have an intrinsic role in the electrical power network. Not only at the generation stations as step-up substations but also at the consumers as step-down substations. These power transformers may be dry or oil-type transformers. In the oil-type power transformers, oil is used as an insulating and cooling medium [
1,
2]. The popular insulating oil used in oil-filled transformers is mineral oil (MO). Mineral oil has significant toxic effects when spilling into the soil and waterways that include the following. (1) MO has low biodegradability with harmful environmental effects. This indicates that just a small portion of oil self-degrades after being released into the environment [
3]. One kilogram of oil leakage waste from a transformer renders 5 million liters of water unfit for consumption [
4]. (2) MO has low flash and fire points. Once a fire hazard occurs, polycyclic aromatic hydrocarbons are released as combustion byproducts, which are toxic byproducts with an extreme threat to the environment [
5]. Further, burning MO releases toxic gases into the environment [
4] and produces heavy and dark smoke. (3) The cost of cleaning up oil spills is often quite high; big environmental utilities spend millions of dollars each year to do so in order to minimize their consequences. (4) MO has inadequate moisture tolerance and poor performance at high temperatures. (5) MO is a non-renewable fossil resource that might run out in the next several decades [
6].
Due to the above-mentioned limitations of MO, great attention must be paid to replacing MO with an alternative oil that permits a higher degree of sustainability, is environmentally friendly, and has the same insulating and cooling properties [
7]. In the last two decades, there has been an ascension in the usage of natural ester oil (NEO) as a strong alternative to mineral oil because of its high biodegradability [
8], which means it easily decomposes into the soil. Hence, many researchers are oriented to study the different properties, characterizations, health index, and condition assessment of natural ester oil using various optical spectroscopy techniques [
9,
10,
11,
12]. Moreover, the NEO properties have been compared to that of MO [
13,
14]. On the other hand, there are some studies interested in the properties of mineral oil mixed with another insulating liquid. Perrier et al. concluded that a mixture of mineral oil with approximately 20% synthetic ester oil achieved an improvement in the dielectric properties and the aging stability without viscosity degradation compared to mineral oil alone [
15]. Nadolny et al. studied the thermal properties of different mixture ratios of mineral and ester oils. According to this study, the optimal ratio of ester oil is 5% at which the heat transferability is highest and thus provides the best cooling performance for the transformer [
16]. Suwarno et al. concluded that an increase in the percentage of ester contents in mineral oil provides an increase in the mixture’s breakdown voltage, although the dissipation factor is slightly degraded [
17]. In [
18], 15% and 20% ester oil ratios provided a better dielectric behavior than other samples in this study. Toudja et al. reported that the measured charging current, resistivity, and mobility of different samples show that a mixture of 85% mineral oil with 15% plant oil is the most excellent mixture [
19]. Dombek et al. presented that, with an increase in the ester contents in synthetic ester/mineral oil mixtures, the flash point and fire point determined with the open cup technique increased [
20]. In [
21], Beroual et al. reported that, under AC or DC applied voltage, Jatropha methyl ester oil has a higher breakdown voltage than MO, and the addition of this oil to MO extensively upgrades the breakdown voltage of the obtained mixture. Further, Ref. [
22] proposes a prediction model of the transformer oil breakdown voltage in the existence of diverse barrier effects for point/plane gap systems with an AC supply voltage using a Box–Behnken design. Recently, Dixit et al. investigated the temperature distribution within the whole structure winding as well as the natural cooling category distribution transformer underneath a retrofilling with natural ester [
23]. In addition, a comprehensive literature review on the mixture of MO with other alternative dielectric fluids such as natural or synthetic esters is presented [
1]. This is helpful for utilities, researchers, as well as transformer owners that are interested in ester liquids besides retrofilling aspects.
As revealed in the above-mentioned literature, most of the existing studies have poor condition assessment and diagnostic tools of the dielectric properties for power transformer oil mixtures under aging processes. To solve this issue, this research work seeks to fill this literature gap by improving the condition assessment and diagnostic performance of the power transformer, specifically the properties of the oil inside that should be enhanced to produce a good insulating and cooling medium under aging. Therefore, replacing or retrofilling aged MO with fresh NEO is one of the approaches to improve power transformer performance. Due to the presence of some mineral oils absorbed by the pressboard between the transformer windings and at the bottom walls of the tank, the oil inside the tank after the replacement process is a mixture of fresh NEO with some quantities of aged mineral oil. The remaining quantities of MO range between 7% and 20% [
1,
24,
25]. In [
24], the breakdown voltage was evaluated for vegetable oil and synthetic oil mixed with MO. When using 20% MO, the drop in the AC breakdown voltage was approximately 3% and 7% for vegetable oil and synthetic oil, respectively. In [
25], the AC breakdown strength of vegetable oils mixed with 10% and 30% MO was investigated. It was found that there is a slight decrement in AC breakdown strength after mixing with mineral oil compared to pure vegetable oil. This decrement attained approximately 6% and 13% for coconut oil and soybean oil, respectively, when mixed with 30% MO. Previous studies on transformer retrofilling considered only the breakdown voltage for investigation. In addition, these studies used MO in the fresh state, which does not represent the actual MO state during the retrofilling process, where MO is usually aged when retrofilling.
Therefore, in this paper, the mixture of NEO with aged MO was investigated to represent the actual condition when retrofilling. Two percentages of aged MO were considered, 10% and 20%, at different aging periods. Two different aging periods were considered, 6 days and 12 days, simulated with an accelerated aging process to be equivalent to 6 years and 12 years in the actual field service. The dielectric properties, breakdown strength, relative permittivity, and dielectric losses were sensed using a LCR meter and oil tester devices for all prepared samples to evaluate the mixtures’ dielectric properties. Also, Weibull distribution analysis was introduced to evaluate the probability of breakdown for the different mixtures. The viscosity of all prepared samples was measured to investigate the dynamic performance of the different mixtures. In addition, the physical mechanism is discussed to present the dielectric and dynamic performance of ester oil mixed with mineral oil in a power transformer. Finally, an electrostatic model of an oil-filled distribution transformer is developed with COMSOL Multiphysics Software to illustrate the electric field distribution inside the transformer in the case of refilling the transformer with natural ester oil (for the best retrofilling oil sample).
4. Physical Mechanisms
Regarding the dielectric strength of the insulating oils, the AC-BDV of insulating oil is mainly related to the relative moisture contents. Therefore, the higher saturation moisture content of mixed insulation oil is a major aspect that produces a higher AC-BDV than that of MO at a similar absolute moisture content [
35,
36]. This concept can be an explanation for the increasing AC-BDV as the percentage of NEO increases in the oil mixture and also with a lower aging period as summarized in
Table 4. The higher aging period of MO produces lower dielectric properties for this oil. This can be attributed to the breaking of some MO molecules due to the applied electric field during the transformer’s continuous operation. From the obtained results, for 20% MO, the aging of MO for 12 days achieved an AC-BDV of 65.3 kV, which is lower than 81.1 kV achieved at 6 days, while for 10% MO, the AC-BDV was reduced from 92 kV to 90.2 kV from 6 days to 12 days aging, respectively. Note that the supply of the measured breakdown voltage here is an AC supply; otherwise, from the literature, the NEO performance under impulse waveforms or other fast transients tends to be lower than that of MO, especially if used in barrier-style insulation or homogenous paper oil insulation. Therefore, it can be interesting to study this performance in future work.
On the other hand, the relative permittivity and the dissipation factor of the mixed insulating oil are not only dependent on the percentage of NEO but also on the aging period as presented in
Table 5. This dependency relates to the dipole polarization inside the mixed oil [
29]. Moreover, the ionic compounds in the aged oil may be dissolved in NEO because of their polar nature; in addition, when the percentage of NEO is increased, this effect becomes prevalent. The increasing dissipation factor of the mixed oil may be due to the aged MO containing a high quantity of polar colloidal components and odorous hydrocarbons [
37].
Regarding the dynamic viscosity of the insulating oils, the lower viscosity of MO for all ranges of temperature refers to a better coolant medium. This behavior may be attributed to its lighter hydrocarbon molecular weight than that inside the mixed oil that contains a large amount of NEO’s heavy triglycerides [
34].
5. Electrostatic Modelling of Oil-Filled Transformer
In this section, an electrostatic model of a 1 MVA, 50 Hz, 22/0.4 kV step-down transformer is simulated using COMSOL Multiphysics Software. The aim of this simulation is to evaluate the distribution of the electrostatic field inside the transformer when using the NEO–MO oil mixtures after the retrofilling process. The case of the oil mixture (10% MO6D + 90% NEO) was considered for the electrostatic field analysis, as it exhibited superior dielectric properties.
COMSOL software is utilized to resolve the desired non-uniform field via the numerical finite element method to effectively model the electrostatic field inside any part of the transformer, specifically the sharp edges and interface points. The input parameters for this model involve the measured r.m.s rates of the AC breakdown voltage followed by the measured dielectric loss and permittivity of the prepared oil samples (NEO–MO mixture). In addition, the other detailed model parameters of the transformer mentioned in
Table 6 and the datasheet parameters of numerous oil types are extracted from real experimental data. The computation of the two-dimensional (2D), axis-symmetric field distribution can be exploited via the finite element method (FEM) through the interface area between the winding and NEO–MO mixture. The mechanism of the COMSOL operation depends on the field region that should first be divided into smaller triangle elements to shrink the energy throughout the wide field area of interest by applying the numerical FEM [
38]. Once a stationary electric field is utilized for a liquid dielectric substance, the Equations (8) and (9) deliver the electrical energy stored inside the full volume of the area under study, taking into consideration the cylindrical model geometry as well as the Laplacian field Equation (8) for static field [
39].
where these equations should be utilized at the unknown potential nodes to compute the electrostatic potential V; after that, the electric field strength E can be easily calculated with Equation (9).
Table 6 presents the technical data for the modeled transformer [
40]. Based on that, the building of an accurate model of the transformer is very difficult, so some simplifications were considered in this work: Only one winding was considered, and the tank in the model was considered to be a perfect cylinder equivalent to one-third of the real transformer tank [
41]. Moreover, an approximately 10% voltage drop across transformer windings was considered due to the copper resistance of these windings that can slightly change the electric field stress along the winding [
40]. Hence, from
Section 2 and
Section 3, all measured parameters of the NEO–MO mixture are used as input data for the simulation model to evaluate the distribution of the electrostatic field inside the transformer when using these oil mixtures.
When COMSOL Multiphysics is started, the model navigator enables the user to begin the modeling process and control all program settings. The space dimension is selected as 2D Axis-symmetric, and the application mode of electrostatic is adopted to begin working on the considered model. The model was created based on the dimensions shown in
Figure 12. Moreover,
Figure 13 presents the model after creation with the space dimension selected as 2D Axis-symmetric. Due to the build of the model under the electrostatic study [
42], the AC voltage distribution inside all parts of the transformer can be obtained as shown in
Figure 14. Furthermore, the electric field distribution can be evaluated based on the FEM using finer grid meshes as presented in
Figure 15. Additionally,
Figure 16 introduces the electric field distribution inside the modeled transformer, specifically at lines AA’ and BB’ and at points a, b, and c (see
Figure 13). From the obtained results, it is clear that the electric field’s highest value points are concentrated between the low-voltage (LV) and high-voltage (HV) windings. These stresses are due to the higher voltage of HV winding compared to LV winding.
To illustrate the values of the electric field in many locations on the insulating oil, the distribution of the electric field is evaluated along the lines AA’ and BB’ presented in
Figure 13. The obtained electric field values along these lines are presented in
Figure 17. Based on the obtained results, along AA’ that lies midway between the HV and LV windings, the higher electric field is presented nearly to the edges of the transformer windings at vertical arc lengths of 310 mm and 905 mm with electric fields of 705 kV/m and 678 kV/m, respectively. On the other hand, along BB’, the higher electric field is presented at a horizontal arc length of 251 mm with an electric field of 706 kV/m. Hence, the highest value points of the electric field on the insulating oil are presented at the edges of the transformer windings due to the direct interface between the transformer winding and the insulating oil molecules in this region. Accordingly, the best insulating oil is (10% MO6D + 90% NEO) from the transformer retrofilling point of view, which produces a good insulation and cooling performance via these highest value points compared to the other oil samples. Moreover, the transformer will have a longer lifetime, better performance, and fewer technical problems. Moreover,
Figure 18 introduces the electric field at points a, b, and c (see
Figure 13) that present a slight variation in the electric field due to the voltage drop along the transformer windings. The obtained results concluded that the electric field was varied at the selected points a, b, and c with values of 615, 636, and 660 kV/m, respectively, which do not exceed 10% variations in the electric field stress along the transformer winding.