Analysis of the Effect of the Degree of Mixing of Synthetic Ester with Mineral Oil as an Impregnating Liquid of NOMEX^® 910 Cellulose–Aramid Insulation on the Time Characteristics of Polarization and Depolarization Currents Using the PDC Method

Abstract

This article continues the authors’ research on NOMEX^® 910 cellulose–aramid insulation saturated with modern electrical insulating liquids, which is increasingly used in the construction of high-power transformers The increase in technical requirements and environmental awareness influences, nowadays, shows that, during the overhaul and modernization of power transformers, petroleum-based mineral oils are increasingly being replaced by biodegradable synthetic esters (oil retrofilling). As a result of this process, the solid insulation of the windings are saturated with an oil–ester liquid mixture with a percentage composition that is difficult to predict. The purpose of the research described in this paper was to test the effect of the degree of mixing of synthetic ester with mineral oil on the diagnostic measurements of NOMEX^® 910 cellulose–aramid insulation realized via the polarization PDC method. Thus, the research conducted included determining the influence of such factors as the degree of mixing of synthetic ester with mineral oil and the measurement temperature on the value of the recorded time courses of the polarization and depolarization current. The final stage of the research involved analyzing the extent to which the aforementioned factors affect parameters characterizing polarization processes in the dielectric, i.e., the dominant dielectric relaxation time constants τ₁ and τ₂, and the activation energy E_A. The test and analysis results described in the paper will allow better interpretation of the results of diagnostic tests of transformers with solid insulation built on NOMEX^® 910 paper, in which mineral oil was replaced with synthetic ester as a result of the upgrade.

1. Introduction

High-power transformers responsible for voltage conversion in distribution and transmission networks are key components ensuring the reliable operation of the power system. Although failures in such equipment are not frequent, they are classified among the most critical fault events due to the complex design of the units, the limited availability of backup transformers, and the high potential repair costs. One of the most severe types of failure that can occur in high-power transformers is an internal short circuit (within the transformer tank), which leads to damage of both the winding and interwinding solid insulation. In many cases, the cost of repairing transformers that have suffered such failures may prove uneconomical compared to purchasing a new transformer unit. Therefore, a key issue from the perspective of ensuring long-term, failure-free operation of these devices is the selection of appropriate materials from which the insulation system will be made. Currently, the vast majority of transformer insulation systems are founded on cellulose-based electrical insulation papers impregnated with mineral oils [1,2,3,4]. Cellulose–oil insulation systems, although widely used, are not without drawbacks. One of the main disadvantages of cellulose is its relatively low thermal resistance, which, at elevated temperatures, leads to the degradation of cellulose fibers. This degradation is manifested, among other things, by the shortening of cellulose macromolecular chains (i.e., a reduction in the degree of polymerization) and the release of water particles as a by-product of cellulose decomposition. Degraded, moisture-saturated insulating paper, in which the aforementioned structural changes have occurred, will exhibit reduced mechanical and electrical strength, which significantly decreases the service life of power transformers.

Due to the emergence of new electrical insulating materials on the market, offering increasingly better electrical and mechanical properties, transformer manufacturers are expanding their portfolios to include transformers built with modern insulation systems, which ensure longer, failure-free operation, even under higher operating temperatures. In the implementation of such projects, electrical insulating papers with a higher thermal class (for example, the cellulose–aramid hybrid insulation NOMEX^® 910) and electrically insulating liquids with high flash points [2,5], among which synthetic esters (e.g., MIDEL^® 7131) are predominant, have proven to be particularly useful. However, it should be noted that the widespread adoption of new insulation systems in the production of high-power transformers depends on multiple factors. In addition to the more obvious considerations, such as the slightly different physicochemical properties of new insulation systems or the costs associated with their use, operational factors must also be taken into account. Among these, one of the most important is the availability of reliable diagnostic tools that enable assessment of the insulation system’s condition during the long-term operation of power transformers. Previous studies, conducted by the authors of this paper [6,7], aimed at evaluating the possibility of adapting one of the widely known and applied diagnostic methods, namely the polarization and depolarization current (PDC) method, to assess the condition of cellulose–aramid hybrid insulation impregnated with a synthetic ester, confirming that the analysis of polarization current I_P and depolarization current I_D can be effectively used to evaluate both the degree of thermal ageing of cellulose fibers and the moisture content of the insulation system.

Synthetic esters, due to their biodegradability [2,8] and very good hygroscopic properties, are increasingly being used both in refurbishment processes (during the drying of cellulose insulation [9,10,11,12,13,14]) and in the modernization of ageing transformer units [12,15,16]. When undertaking such procedures, it must be understood that in power transformers with insulation systems based on cellulose, complete replacement of the insulating liquid with a new one is not possible. In a transformer where the insulating liquid has been replaced from a mineral oil to a synthetic ester, the solid winding insulation will remain impregnated with an insulating liquid that is a mixture of the residual mineral oil and the newly introduced synthetic ester [12,15,16,17,18]. Accordingly, in order to complement previous research on the adaptation of the PDC method for the diagnostics of cellulose–aramid-oil insulation, this paper presents the results of a study illustrating the effect of the degree of mixing between mineral oil and synthetic ester on the values of the recorded polarization current I_P and depolarization current I_D, as obtained using the PDC method. This study also presents the results of analyses demonstrating the influence of the aforementioned factor on the value of activation energy E_k and the dominant relaxation time constants τ₁ and τ₂, (which are associated with polarization processes occurring in the respective layers of the cellulose–aramid hybrid insulation). These quantities made it possible to determine the effect of the impregnation mixture used on the polarization processes occurring in the insulating system under study, which can be helpful in determining the composition of the insulating system mixture.

The results of the experiments made it possible to demonstrate the sensitivity of the PDC method to the percentage composition of the mixture of the two liquids, as manifested by the change in the values of the recorded time courses of I_P polarization and I_D depolarization currents. The results obtained as a result of laboratory experiments also made it possible to determine the potential operational hazards that may arise as a result of operating the cellulose–aramid–oil–ester insulation system.

2. Materials Used in the Laboratory Tests

2.1. NOMEX^® 910 Cellulose–Aramid Electrical Paper

For the laboratory tests, cellulose–aramid electrical insulating paper NOMEX^® 910 manufactured by DuPont™ was used. NOMEX^® 910 insulation combines the best properties of widely used cellulose-based Kraft insulation with DuPont’s proprietary aramid synthetic polymer, commercially known as NOMEX^®. The NOMEX^® 910 electrical insulating material features a layered structure consisting of three layers, in which the base of the NOMEX^® 910 paper is formed from high-quality thermally upgraded cellulose pulp. Once formed into a paper sheet, it is coated on both sides with a thin layer of high-temperature meta-aramid binder NOMEX^®. The structure of the cellulose–aramid hybrid insulation NOMEX^® 910, along with images of the sheet, and an example of its application, are shown in Figure 1 and Figure 2. The development of electrical insulating paper based on this structure has resulted in a unique hybrid that combines the most advantageous properties of both cellulose-based and aramid-based insulation.

The combination of two insulating materials within the structure of NOMEX^® 910 insulating paper has enabled improved absorption capacity for insulating liquids compared to single-layer aramid-based NOMEX^® material, as well as an increased thermal class of the resulting material in comparison to the commonly used cellulose-based Kraft insulation in transformer manufacturing. When impregnated with mineral oil, NOMEX^® 910 paper exhibits a thermal class of 130 °C. For comparison, conventional cellulose-based Kraft insulating papers, when impregnated with the same insulating liquid, typically have a thermal class of 110–120 °C. The thermal class of NOMEX^® 910 paper can be further increased to 140 °C by impregnating the insulation with esters.

Table 1. Typical mechanical and electrical properties of NOMEX^® 910 paper [19].

Property	Units	Value	Test Method
Basis weight	g/m2	80	ASTM D646
Burst strength	N/cm2	27	ASTM D828
Tensile strength, MD 1	N/cm	70	ASTM D828
Tensile strength, XD 2	N/cm	17	ASTM D828
Elongation, MD	%	2.2	ASTM D828
Elongation, XD	%	6.9	ASTM D828
Tear strength, MD	N	0.45	TAPPI 414
Tear strength, XD	N	0.70	TAPPI 414
AC rapid rise breakdown in Ester Liquid	kV/mm	87	ASTM D149
Dielectric Constant at 60 Hz, 23 °C, Ester Liquid	-	4.2	ASTM D150
Dissipation Factor at 60 Hz, 23 °C, Ester Liquid	%	0.9	ASTM D150

¹ Machine direction; ² Cross machine direction.

presents the technical specification of NOMEX^® 910 paper with a thickness of 0.08 mm.

Taking into account the data presented in Table 1, it can be stated that NOMEX^® 910 insulation demonstrates very good electrical and mechanical properties. NOMEX^® 910 exhibits higher mechanical strength in the machine direction (MD) compared to the cross direction (XD). The moderate flexibility of this material allows for easy forming, which is particularly important when constructing insulation systems with complex geometries. In addition, according to the manufacturer’s data [19], NOMEX^® 910 paper is characterized by stable dielectric properties across a wide range of operating temperatures and moisture levels. Due to its properties, NOMEX^® 910 paper is intended for use as an electrical insulating material in power equipment requiring high dielectric strength and resistance to elevated operating temperatures, particularly in the construction of insulation systems for large power transformers, as well as for motors and generators.

2.2. Impregnation Liquids: Synthetic Ester MIDEL^® 7131 and Mineral Oil NYTRO 10 XN

During laboratory tests, a mixture of two insulating liquids—mineral oil NYTRO 10 XN and synthetic ester MIDEL^® 7131—was used to impregnate NOMEX^® 910 paper. Mineral oils are the most commonly used dielectric liquids for the impregnation of insulation systems based on cellulose. Due to their good electrical insulating properties and relatively low cost (compared, for example, to esters), they remain the dominant liquids on the market for the production of large power transformers [2,12,20]. The mineral oil NYTRO 10XN used during the laboratory tests is a liquid produced by the Swedish company Nynas AB and, like all mineral oils, is a product of crude oil refining and, therefore, non-biodegradable. Due to its high resistance to oxidation, this oil is widely used as an impregnating liquid in power transformers.

Table 2. Typical physical, chemical, and electrical properties of mineral oil NYTRO 10XN [21].

Property	Units	Value	Test Method
Density, 20 °C	kg/dm3	0.895	ISO 12185
Viscosity, 40 °C	mm2/s	8.0	ISO 3104
Viscosity, −30 °C	mm2/s	800	ISO 3104
Fresh point	°C	140	ISO 2719
Biodegradation	Non-Biodegradable
Interfacial tension	mN/m	>40	EN 14210
Pour point	°C	−45	ISO 3016
Water content	mg/kg	30	IEC 60814
Neutralization value	mg KOH/g	<0.01	IEC 62021
Power factor at 90 °C	-	<0.005	IEC 60247
Breakdown voltage
-before treatment -after treatment	kV kV	>30 >70	IEC 60156 IEC 60296

presents the technical data of the NYTRO 10XN mineral oil.

The second impregnating liquid used during the laboratory tests was the synthetic ester MIDEL^® 7131, produced by MIDEL & MIVOLT Fluids Ltd. As a fully biodegradable liquid, this ester serves as an excellent substitute for commonly used petroleum-based mineral oils. According to the manufacturer’s technical data, MIDEL^® 7131 can be successfully used as an insulating liquid in equipment with voltages up to 433 kV. It exhibits impressive water absorption capacity (up to 600 ppm) [12] without a reduction in breakdown voltage, and a very high saturation limit (up to 2700 ppm at 20 °C), which prevents the precipitation of water. In addition, this liquid allows for an increase in operating temperature without reducing the service life of the insulation system.

Table 3. Typical physical, chemical, and electrical properties of ester MIDEL^® 7131 [22].

Property	Units	Value	Test Method
Density, 20 °C	kg/dm3	0.97	ISO 3675
Viscosity, 40 °C	mm2/s	29	ISO 3104
Viscosity, −30 °C	mm2/s	1440	ISO 3104
Fresh point	°C	260	ISO 2719
Biodegradation	Readily Biodegradable
Fire point	°C	316	ISO 2592
Pour point	°C	−56	ISO 3016
Water content	mg/kg	50	IEC 60814
Crystallization	-	No crystals	IEC 61099
Power factor at 90 °C	-	<0.008	IEC 60247
Dielectric Breakdown	kV	>75	IEC 60156
DC Resistivity at 90 °C	GΩ·m	>20	IEC 60247

presents selected physicochemical properties of the MIDEL^® 7131 ester.

Synthetic esters (including MIDEL^® 7131), as previously mentioned, are also used in the modernization of ageing transformer units that have been filled with mineral oil for many years. In this context, the synthetic ester can be used in two ways: either as a new filling liquid for the transformer tank, or as a medium applied during refurbishment processes, such as the replacement of the dielectric liquid filling the tank or the regeneration (drying) of the transformer’s solid cellulose insulation. When carrying out such procedures, it should be remembered that the assessment of the condition of this type of insulation system (i.e., cellulose and a mixture of mineral oil and synthetic ester) will only be possible once physicochemical equilibrium has been achieved inside the transformer, resulting from the complete mixing of both dielectric liquids within the tank.

3. PDC Method

The PDC (polarization and depolarization current) method was used during the laboratory tests. It is one of the polarization-based diagnostic methods applied to assess the degree of thermal ageing and moisture content in the insulation systems of power equipment, including power transformers. The principle of the method is based on the phenomenon of dielectric polarization, which occurs when a dielectric material is placed in a direct current electric field. The PDC method is based on the analysis of the time-dependent behavior of the dielectric’s polarization current I_P and depolarization current I_D. This method is classified as an off-line diagnostic technique, meaning it is carried out on equipment disconnected from the power supply network. Figure 3 shows a schematic diagram of the test setup used for examining cellulose–aramid–oil insulation samples using the PDC method. The setup was designed to replicate, as accurately as possible, the conditions under which measurements are typically performed on power transformers in service—most often at transformer substations. Figure 4 presents the time characteristics of currents and voltages recorded during diagnostic measurements using the PDC method.

The concept of performing diagnostic measurements on insulation samples using the test setup shown in Figure 3 is based on applying a direct voltage source U_C (set to 500 V during the experiments) to the test object (with relay P₁ in the closed position) and measuring and recording the resulting dielectric polarization current I_P over a period t_P (600 s during the measurements). As shown in the time characteristics in Figure 4, the value of the recorded polarization current I_P decreases over time until it stabilizes at the level of the conduction current I_K, which results from the finite transverse resistivity of the tested insulation sample. After the time t_P has elapsed, the voltage source is disconnected, and the terminals of the test object are short-circuited (the control system deactivates relay P₁ and activates relay P₂). From this moment, the depolarization current I_D of the dielectric is measured and recorded over a period t_D (also set to 600 s during the measurements). It can then be observed that a depolarization current I_D flows in the circuit, with its magnitude decreasing over the time t_D until it reaches zero, indicating complete discharge of the dielectric. The end of the depolarization period t_D is determined by the deactivation of relay P₂.

As mentioned in the introduction to this paper, the recorded time characteristics of the polarization current I_P and depolarization current I_D were subsequently used in the analysis to determine parameters characterizing the polarization process of the tested dielectric, namely the activation energy E_A and the dominant relaxation time constants τ₁ and τ₂.

The polarization current I_P waveforms obtained during the laboratory experiments were used to determine the activation energy E_A, i.e., the minimum amount of energy required to initiate the reorientation of electric dipoles in the dielectric as a result of the application of an external electric field. The activation energy E_A of individual test samples was determined using the regression function of the polarization current I_P(t), expressed in the form of the Jonscher LFD equation, given as Equation (1) [6,7,23]:

$$I_P\left(t\right)\propto A_1·t^{{-N}_1}+A_2·t^{{-N}_2},$$

(1)

where A₁, A₂, N₁, N₂ are function parameters.

By using the parameters of the polarization current regression function and applying linear approximation to the temperature-dependent Arrhenius plot, the activation energy E_A of the sample can be calculated using Equations (2)–(4):

$$t_A=\sqrt[{-N}_1{+N}_2]{{\displaystyle \frac{A_2}{A_1}}},$$

(2)

$$\ln (t_A)=f\left({\displaystyle \frac{1000}{T}}\right),$$

(3)

$$E_A=1000·a·k,$$

(4)

where t_A—characteristic time (followed by a change in the slow relaxation process); T—temperature of the sample (in degrees Kelvin), E_A—activation energy (eV); a—directional coefficient of the linear regression function, k—Boltzmann constant.

The depolarization current waveforms I_D(t), analyzed from the measurement results, were used to determine the dominant relaxation time constants—i.e., the times after which polarization mechanisms reach equilibrium in response to a change in the electric field. Considering that the NOMEX^® 910 material has a layered cellulose–aramid structure, the depolarization current regression function was based on the Debye equation with two relaxation time constants. The relaxation time constant τ₁ is correlated with the relaxation process of the aramid fibers, while the time constant τ₂ is associated with the relaxation of the cellulose fibers.

$$I_D\left(t\right)\propto B_1·e^{-{\displaystyle \frac{t}{{\tau }_1}}}+B_2·e^{-{\displaystyle \frac{t}{{\tau }_2}}},$$

(5)

where B₁, B₂—parameters of the function; τ₁, τ₂—dominant time constants of the relaxation processes of aramid and cellulose fibers, respectively.

4. Sample Preparation Method

4.1. Metering Samples Preparation

For the purpose of conducting laboratory tests analyzing the influence of the mixing ratio of mineral oil and synthetic ester, used as the impregnating liquid for cellulose–aramid NOMEX^® 910 insulation, on the polarization and depolarization current characteristics obtained using the PDC method, five containers with a capacity of 2 liters were prepared and filled with dielectric liquids in the following proportions:

• 100% mineral oil NYTRO 10XN
• 75% mineral oil NYTRO 10XN + 25% synthetic ester MIDEL^® 7131
• 50% mineral oil NYTRO 10XN + 50% synthetic ester MIDEL^® 7131
• 25% mineral oil NYTRO 10XN + 75% synthetic ester MIDEL^® 7131
• 100% synthetic ester MIDEL^® 7131

The proportions of the individual liquids in the impregnation jars determined the analyzed influence of the mixing ratio of synthetic ester and mineral oil on the recorded polarization I_P and depolarization I_D current waveforms of the insulation system during the measurements. Figure 5 shows the dielectric liquids prepared for the impregnation process of NOMEX^® 910 paper.

In parallel with the process described above, NOMEX^® 910 insulation samples were prepared for the impregnation procedure. The preparation of laboratory samples involved cutting sheets of NOMEX^® 910 paper into samples with dimensions of 100 × 1300 mm. The number of samples cut corresponded to the number of tested mixing ratios of mineral oil and synthetic ester. Next, the NOMEX^® 910 insulation samples underwent a drying process at a temperature of 110 °C for a duration of 25 h. The selected drying temperature of 110 °C ensures the evaporation of moisture contained within the structure of the NOMEX^® 910 paper, without initiating ageing processes, due to the high thermal class of the tested paper. The drying process of the test samples was carried out using a high-temperature laboratory oven, which enabled the set heating temperature to be maintained for any desired duration. A photograph of the laboratory oven with NOMEX^® 910 paper samples prepared for drying is shown in Figure 6a. The next stage in sample preparation involved drying the samples for an additional 25 h under laboratory vacuum conditions at a pressure of approximately 10 Pa (Figure 6b).

The next step in the preparation of the test samples was their impregnation with mixtures of dielectric liquids. For this purpose, the dried NOMEX^® 910 insulation samples were immersed in mixtures of liquids preheated to 70 °C and subjected to a drying and degassing process under laboratory vacuum conditions at a pressure of approximately 10 Pa for a duration of 25 h. The samples prepared in this way were then subjected to a prolonged conditioning process in tightly sealed containers (without air access).

4.2. Measurement System

To replicate the layered nature of the insulation system in power transformers, the test samples with dimensions of 100 × 1300 mm were wound onto a brass measuring rod (160 mm in length and 40 mm in diameter), which served as the low-voltage electrode. The circumference of the measuring rod and the dimensions of the test sample allowed for the application of 10 layers of insulation. An aluminum foil strip, 80 mm wide wound onto the test sample, served as the high-voltage electrode. Figure 7 shows a cross-section of the electrode arrangement on the measuring rod. The design of the measuring rod was developed to allow for temperature regulation of the insulation sample during measurements, thereby replicating the varying thermal conditions to which the transformer insulation system is subjected as a result of the changing load. For this purpose, two holes were drilled into the rod structure to accommodate the installation of a heating element and a PT 100 temperature sensor. The PT 100 sensor enabled temperature control of the measuring rod and was connected to a controller that regulated the operation of the heater. To ensure stability of the measurement setup during the experiments, the measuring rod was mounted on two Teflon stabilizers, which also provided galvanic isolation of the electrode system from the surroundings. Figure 8 shows a cross-section of the insulation sample wound onto the measuring shaft.

Figure 9 presents photographs of the measuring rod and the NOMEX^® 910 insulation sample prepared for measurement, both before and after the electrode system was assembled. To secure the insulation sample to the measuring rod during the winding process, a longitudinal groove with a depth of 10 mm was cut along the entire length of the rod. Additionally, to ensure tight winding of the sample onto the rod, auxiliary cable ties made of non-conductive polyamide 6.6 were used during assembly. The shielding (grounding) electrode was made from stainless steel hose clamps.

To maintain uniform thermal conditions during measurements—i.e., consistent temperatures of the measuring rod and the surrounding environment—the test stand was equipped with a temperature stabilizer, ensuring stable thermal conditions with an accuracy of ±1 °C. The PDC measurements of NOMEX^® 910 insulation were carried out using the SONEL MIC-15k1 insulation resistance meter. This device enabled the measurement and recording of the dielectric polarization and depolarization currents based on its built-in function for measuring the dielectric discharge (DD) factor. Figure 10 shows a photograph of the test setup used for insulation diagnostics with the PDC method.

5. Experimental Results

For all measurements of polarization and depolarization currents, a uniform data recording time was used, with t_P = t_D = 600 s. The currents were recorded by the SONEL MIC-15k1 meter with a sampling frequency of approximately 2 Hz. Due to the fact that the meter was measuring very low current values, to eliminate potential interference, the operation of the device was controlled remotely via Bluetooth^® using the Sonel MIC Mobile application.

5.1. Effect of Temperature

A key factor directly influencing the values of the polarization and depolarization currents recorded using the PDC method is the measurement temperature. During the laboratory experiments, measurements were conducted at temperatures ranging from 30 °C to 70 °C, with increments of 10 °C. The selected temperature range corresponds to the operating temperatures that transformer insulation may be exposed to under real conditions, where 30 °C represents the insulation temperature after a prolonged outage. The temperature of 70 °C corresponds to the insulation temperature of a transformer that was operating at rated load just before being switched off.

In the analysis of the influence of measurement temperature on the values of the recorded polarization I_P and depolarization I_D current waveforms, the focus was limited to NOMEX^® 910 insulation samples impregnated with a mixture of two dielectric liquids in a 75% mineral oil and 25% synthetic ester ratio. This choice was made because similar trends were observed for the other liquid mixtures. Figure 11 and Figure 12 show the time-dependent waveforms of polarization current I_P and depolarization current I_D, respectively, illustrating the effect of measurement temperature on the recorded current values for unaged, dry cellulose–aramid NOMEX^® 910 insulation impregnated with the above-mentioned liquid mixture. For comparison purposes, Figure 13 shows the influence of measurement temperature on the time-dependent depolarization current I_D waveforms measured on an unaged, dry NOMEX^® 910 paper sample impregnated with pure synthetic ester.

Based on the time characteristics presented in Figure 11, it was demonstrated that the measurement temperature affects the values of the recorded polarization currents flowing through NOMEX^® 910 insulation impregnated with a liquid mixture in the ratio of 75% mineral oil and 25% synthetic ester. It was shown that a 10 °C increase in measurement temperature results in an approximately proportional increase in the polarization current IP values across the entire recording time range. This increase in current values is attributed to the decreasing resistivity of the impregnating liquids as well as the cellulose component of the NOMEX^® 910 paper structure saturated with the mixture of both dielectric liquids. Previous studies conducted by the authors of this paper indicate that this trend is also observed for pure mineral oils and synthetic esters [6,24].

Referring to the time-dependent depolarization current characteristics in Figure 12, it was demonstrated that the higher the sample temperature, the more rapid was the decrease in the depolarization current values. This trend results from the fact that an increase in sample temperature leads to a reduction in the viscosity of the insulating liquid used to impregnate the NOMEX^® 910 insulation. In addition, an increase in temperature leads to intensified vibrations of the electric dipoles within the tested insulation, which facilitates the process of returning to the original disordered state of the dipoles—i.e., the depolarization process.

By analyzing the initial stage of the depolarization process (t < 10 s, i.e., a few seconds after short-circuiting the terminals), it can be observed that samples at higher temperatures exhibit lower recorded depolarization current I_D values throughout the entire recording period. This trend was observed in all NOMEX^® 910 paper samples impregnated with mixtures of mineral oil and synthetic ester, specifically in the proportions of 25%/75%, 50%/50%, and 75%/25%. The time-dependent depolarization current I_D waveforms measured on samples impregnated with these mixtures, during the initial phase of the depolarization process (t < 10 s), therefore, exhibit different behavior compared to samples impregnated with pure dielectric liquids—such as mineral oils [24], synthetic esters [6], or natural esters [25]—where, in the first seconds after short-circuiting the current terminals, samples at higher temperatures display higher recorded depolarization current values (see Figure 13). This directly results from the charge state of the sample just before the current terminals are short-circuited. This phenomenon is therefore not observed in samples impregnated with mixtures of mineral oil and synthetic ester. Consequently (as shown in Figure 11 and Figure 12), a situation occurs in which a sample exhibiting a higher polarization current I_P intensity (compared to measurements taken at lower temperatures) also exhibits a lower depolarization current I_D intensity throughout the entire recording period. A charge deficit is therefore observed between the polarization and depolarization processes of the tested insulation system. One explanation for this observation (i.e., the difference in free charge resulting from the asymmetry between the polarization and depolarization processes) may be the phenomenon of local electric charge trapping [12,26]. This effect involves the accumulation of free charges in local energy minima, i.e., regions within the non-homogeneous structure of the insulation system (such as phase boundaries), where differences in electrical conductivity (σ) and permittivity (ε) of the insulation components occur [12,27]. Another possible explanation for the observed phenomenon may be the discharge of part of the charge outside the detection window during the initial microseconds of the depolarization process, which were not recorded. Considering the fact that the analyzed insulation system consists of multiple components—i.e., it is a cellulose–aramid–oil–ester insulation—it appears likely that some free charges remained trapped within the structure of the insulation system, despite the depolarization process of the dielectric. It should therefore be noted that the tested insulation system may exhibit free charge accumulation at phase boundaries, which over time or under the influence of elevated temperatures can lead to the occurrence of internal partial discharges. If these discharges develop further, they may lead to complete breakdowns and transformer failure [28].

5.2. Analysis of the Influence of Mixing Mineral Oil and Synthetic Ester

Figure 14 and Figure 15 present the time-dependent waveforms of polarization current I_P and depolarization current I_D measured on unaged, dry NOMEX^® 910 insulation samples impregnated as follows:

• Sample no. 1—100% mineral oil NYTRO 10XN
• Sample no. 2—75% mineral oil NYTRO 10XN + 25% synthetic ester MIDEL^® 7131
• Sample no. 3—50% mineral oil NYTRO 10XN + 50% synthetic ester MIDEL^® 7131
• Sample no. 4—25% mineral oil NYTRO 10XN + 75% synthetic ester MIDEL^® 7131
• Sample no. 5—100% synthetic ester MIDEL^® 7131

The characteristics presented in Figure 14 and Figure 15 refer to measurements carried out at a representative temperature of 40 °C; similar trends were observed at other measurement temperatures.

The time-dependent polarization current I_P curves shown in Figure 14 confirmed that pure mineral oil has significantly higher resistivity compared to pure synthetic ester. Mixing mineral oil with synthetic ester results in a non-linear (disproportionate to concentration) decrease in the resistivity of the insulation system. Analysis of the polarization currents measured on NOMEX^® 910 insulation samples impregnated with mineral oil–synthetic ester mixtures indicated the following:

• The 25% mineral oil–75% synthetic ester concentration exhibits dominant characteristics of the synthetic ester, meaning the resulting mixture behaves like the synthetic ester in terms of its electrical properties. As a highly polar liquid, the synthetic ester—under the influence of the polarizing factor (electric field intensity)—may cause separation of mineral oil molecules, which are non-polar in nature [12,29].
• In the case of the 75% mineral oil–25% synthetic ester mixture, the flow of polarization current I_P results from the movement of charges (ions) and the dipole polarization of particles. Although the synthetic ester constitutes a smaller proportion than the mineral oil, its more polar structure introduces polar molecules into the system, ultimately increasing dipole polarization. As the dominant liquid, mineral oil may facilitate ion movement, as it does not absorb them as effectively as pure synthetic ester, thus acting as a transport medium.

Therefore, synthetic ester dispersed in mineral oil provides polarization but does not form a compact structure that could restrict ion mobility.

• In the case of a 50% mineral oil–50% synthetic ester mixture, the presence of a larger amount of synthetic ester leads to the formation of more complex molecular structures, which limits ion mobility and reduces the polarization current of the insulation system.

The depolarization current I_D characteristics provide insight into the relaxation processes within the dielectric system, which depend on the molecular structure and the associated dielectric polarization phenomena.

Referring to the time-dependent depolarization current characteristics shown in Figure 15, the following observations were made:

• Homogeneous liquids exhibit the extreme values of the recorded depolarization currents.
• Mineral oil, used as the impregnation liquid for NOMEX^® 910 insulation, is a non-polar liquid characterized by a low contribution of relaxation mechanisms and minimal polarizability, resulting in the lowest recorded depolarization current.
• NOMEX^® 910 paper impregnated with pure synthetic ester showed the highest depolarization current levels, indicating strongly pronounced relaxation processes due to the structure of the synthetic ester and its highly polar nature compared to the other insulating liquid mixtures used in the laboratory tests.
• The time-dependent depolarization current characteristics of samples impregnated with liquid mixtures (mineral oil–synthetic ester) lie between the values recorded for samples impregnated with homogeneous insulating liquids. The depolarization current increases with a higher proportion of synthetic ester in the mixture.
• In the 75% mineral oil–25% synthetic ester mixture, the small amount of synthetic ester introduces additional relaxation mechanisms due to its polarizability, though not enough to significantly reduce the mixture’s resistivity.
• The 75% mineral oil–25% synthetic ester mixture exhibited the highest depolarization current decay rate, which may be attributed to the fact that the depolarization current is the sum of two components: ion movement and the relaxation of electric dipoles. The dominance of mineral oil (which has lower viscosity than synthetic ester) creates an environment favorable for ion mobility.

6. Analysis Results

6.1. Influence of the Mixing Ratio of Mineral Oil and Synthetic Ester on Activation Energy E_A

Figure 16 presents the relationship between activation energy E_A and the mixing ratio of mineral oil NYTRO 10XN with synthetic ester MIDEL^® 7131, used as the impregnation liquid for cellulose–aramid NOMEX^® 910 insulation.

The characteristic shown in Figure 16 revealed an approximately linear trend in the changes of the determined activation energy E_A; i.e., the activation energy increases with the rising concentration of synthetic ester in the analyzed liquid mixture. The observed trend can be explained as follows:

• The lower activation energy E_A of the sample impregnated with pure mineral oil, compared to the sample impregnated with pure synthetic ester, results from the non-polar nature of NYTRO 10XN oil. The polarization processes occurring in mineral oil are due to the induction of temporary dipoles by the electric field. Such polarization requires less energy to disturb the symmetric electron clouds of the non-polar mineral oil molecules. Mineral oil molecules are structurally simple and flexible, making the formation of electron clouds less energy-intensive than the polarization process of synthetic ester molecules, which are characterized by a much more complex structure.
• The higher activation energy E_A observed for the sample impregnated with synthetic ester, compared to the one impregnated with pure mineral oil, results from the polar nature of MIDEL^® 7131 ester, whose molecular structure includes permanent dipole moments. The polarization process of such a dielectric requires an energy-intensive effort to overcome internal molecular interactions (such as hydrogen bonding and dipole orientation within the liquid) in order to align the dipoles in the electric field.
• The higher viscosity of the synthetic ester MIDEL^® 7131 compared to the mineral oil NYTRO 10XN [12,30,31] affects the ability of liquid molecules to orient themselves in an electric field. Higher viscosity implies greater resistance to molecular movement and, consequently, a higher energy requirement.
• An increase in the concentration of synthetic ester in the liquid mixture leads to an increase in the activation energy E_A required to polarize the dielectric. This pattern results from the increased viscosity of the liquid and the higher degree of polarity in the liquid containing permanent dipole moments.

6.2. Influence of the Mixing Ratio of Mineral Oil and Synthetic Ester on the Dominant Relaxation Time Constants τ₁ and τ₂

Figure 17 and Figure 18 present the influence of the mixing ratio of mineral oil NYTRO 10XN with synthetic ester MIDEL^® 7131, used as the impregnation liquid for NOMEX^® 910 insulation, on the values of the dominant relaxation time constants τ₁, associated with the relaxation processes of aramid fibers, and τ₂, correlated with the relaxation processes of cellulose fibers.

Analysis of the influence of measurement temperature on the values of the determined dominant relaxation time constants τ₁ and τ₂ revealed a decrease in both time constants with increasing sample temperature. An increase in temperature enhances molecular mobility in liquids, which directly leads to a reduction in the relaxation times τ₁ and τ₂.

The characteristics shown in Figure 17, presenting the influence of the mixing ratio of mineral oil and synthetic ester on the value of the dominant relaxation time constant τ₁, demonstrated that homogeneous liquids used to impregnate the cellulose–aramid NOMEX^® 910 insulation exhibit extreme values of the determined τ₁ constants. Samples impregnated with the synthetic ester MIDEL^® 7131, at corresponding measurement temperatures, display longer electric dipole relaxation times (approximately 20 s) compared to those impregnated with the mineral oil NYTRO 10XN.

The values of the determined dominant relaxation time constants τ₁ for the liquid mixtures, at corresponding measurement temperatures, were found to lie between the extreme values defined by the homogeneous liquids. Moreover, with increasing concentration of synthetic ester in the mixture, the τ₁ time required to reach dipole equilibrium in the tested insulation system increased.

The explanation for these relationships can be attributed to the following factors:

• The polarization of NYTRO 10XN mineral oil, being a non-polar liquid, occurs through the induction of temporary electric dipoles under the influence of an external electric field. This process is relatively fast, as mineral oil molecules do not possess permanent dipole moments.
• Synthetic esters (including MIDEL^® 7131) are polar liquids, meaning that their molecules contain permanent dipole moments, and their polarization (i.e., reorientation in response to an applied electric field) occurs through the overcoming of intermolecular interactions. This means that the dipoles of the synthetic ester molecules require more time to align in the electric field due to resistance caused by internal molecular interactions.
• The increase in relaxation time τ₁, with rising synthetic ester concentration in the liquid mixture, results from the growing number of molecules containing permanent dipole moments.

Referring to the characteristics shown in Figure 18, which present the influence of the mixing ratio of mineral oil and synthetic ester on the dominant relaxation time constant τ₂ of cellulose fibers, the following observations were made:

• Pure synthetic ester resulted in the longest relaxation time of cellulose fibers, which is attributed to the highest polarity of the tested insulation structure.
• The 50% mineral oil–50% synthetic ester mixture exhibited intermediate properties between the non-polar pure mineral oil NYTRO 10XN and the polar synthetic ester MIDEL^® 7131. The electric dipoles interact more strongly with the synthetic ester molecules, though not to the same extent as in the case of pure ester.
• Pure mineral oil showed a shorter τ₂ relaxation time than both the pure ester and the 50%/50% mixture, which can be explained by its non-polar nature and the mechanism of polarization, which proceeds much faster compared to the polarization of molecules with permanent dipole moments.

The τ₂ values of samples impregnated with mixtures in the 25%/75% and 75%/25% proportions appeared to be the most surprising. Despite the increased concentration of synthetic ester, the determined relaxation times τ₂ for cellulose fibers turned out to be lower than that of the sample impregnated with pure mineral oil. The lowest τ₂ value (for corresponding measurement temperatures) was obtained for the 75% mineral oil/25% synthetic ester mixture. This observation is likely caused by multiple complex factors, among which the following are significant:

• Mixing mineral oil and synthetic ester in the proportions described above may lead to the formation of a molecular environment with unique properties that alter the nature of interactions with cellulose fibers.
• A reduction in the viscosity of the resulting liquid mixture may occur, as mineral oil molecules can act as a diluent, which may in turn reduce long-lasting hydrogen bonding between the liquid and cellulose.
• In unbalanced proportions of mineral oil and synthetic ester, unique interactions may arise that modify the interaction between the liquid and cellulose, which, through synergistic effects, contribute to a more dynamic relaxation process of the cellulose fibers.

7. Conclusions

The authors’ intention was to present the influence of a proportional increase in the concentration of synthetic ester in a mixture with mineral oil, in comparison to pure insulating liquids, which were the subject of previous studies [6,24]. Therefore, to investigate the effect of the mixing ratio of mineral oil and synthetic ester, mixtures were prepared in the following proportions: 25%/75%, 50%/50%, and 75%/25%, and used as impregnation liquids for cellulose–aramid NOMEX^® 910 insulation. Subsequently, the nature of the changes in the PDC characteristics and the parameters describing the polarization processes in the tested insulation system were analyzed.

The laboratory measurements carried out using the PDC method, along with the subsequent analysis of the parameters characterizing the polarization process of the tested insulation system, confirmed that the mixing ratio of mineral oil and synthetic ester—used as the impregnation medium for cellulose–aramid NOMEX^® 910 insulation—has an influence on both the time-dependent characteristics of the polarization current I_P and depolarization current I_D, as well as on the key parameters of the polarization process, namely the activation energy E_A and the dominant relaxation time constants τ₁ and τ₂.

As in the case of homogeneous electrical insulating liquids (mineral oil, synthetic ester, natural ester) used as impregnating liquid of insulation NOMEX^® 910, an increase of 10 °C in the measurement temperature causes an approximately proportional increase in the value of I_P polarization currents, over the entire range of the adopted recording time. This relationship is characteristic of insulating systems built on cellulose.

A higher measurement temperature translating into an increase in the value of the polarization current does not translate into a higher initial value of the depolarization current, as is the case with homogeneous dielectric liquids, i.e., mineral oil and synthetic and natural esters used to impregnate cellulose–aramide insulation NOMEX^® 910. This may cause the insulating system based on NOMEX^® 910 paper impregnated with a mixture of mineral oil–synthetic ester liquid to be characterized by the accumulation of free charges at the phase boundary, which, with time or under the influence of high temperature, may cause the occurrence of internal incomplete discharges. With time, this may lead to the formation of complete discharges and failure of the transformer.

Mixing mineral oil with synthetic ester causes a disproportionate (to the concentration of the mixture) reduction in the resistivity of the insulating system, which translates directly into the value of the recorded time courses of I_P polarization and I_D depolarization currents.

Undoubtedly, the most useful tool for assessing the degree of liquid mixing proved to be the proportional correlation of the activation energy E_A (Figure 16). Naturally, determining this parameter requires measurements to be taken at multiple insulation temperatures, which can be challenging in on-site conditions. However, thanks to the off-line nature of the PDC method, it is feasible to perform repeated measurements during the natural cooling process of a transformer that has been disconnected from the grid. Given that the degree of mixing of mineral oil and synthetic ester affects the value of the recorded time courses of the I_P polarization current and I_D depolarization of the tested dielectric (according to the relationships described in Section 5.2), the determination of the activation energy of E_A can be a useful tool for determining the degree of mixing of mineral oil and synthetic ester in the impregnation liquid used. This will allow a more reliable interpretation of the diagnostic measurements of the transformer obtained using the PDC method.

Given that, in the case of replacing mineral oil with synthetic ester in a modernized transformer unit, the mineral oil—after equilibrium is reached through mixing—will ultimately constitute a minority component of the impregnation liquid for the transformer’s solid insulation, the results and analyses presented in this study should be repeated for mixtures with a majority share of synthetic ester (e.g.,: 5%/95%, 10%/90%, 15%/85%). However, the initial stage of mineral oil migration into the synthetic ester from the inner layers of the transformer’s solid insulation may be sufficiently prolonged for the presented research results to remain useful in assessing the degree of liquid replacement within the solid insulation of the transformer.