Thermal and Electrical Fault Diagnosis in Oil–Paper Insulation System: A Comparative Study of Natural Esters and Mineral Oil

Abstract

Power transformer insulation systems, composed of liquid and solid insulators, are continuously exposed to thermal and electrical stresses that degrade their performance over time and may lead to premature failure. Since these stresses are unavoidable during operation, selecting effective insulating materials is critical for long-term reliability. In this study, Kraft insulation paper was used as the solid insulator and impregnated with three different liquids: mineral oil and two natural esters (NE1204 and NE1215), to evaluate their stability under simultaneous thermal and electrical stress. The degradation behavior of the oil-impregnated papers was assessed using frequency-domain dielectric spectroscopy (FDS) and Fourier-transform infrared spectroscopy (FTIR), enabling early fault detection. Comparative analyses were conducted to evaluate the withstand capability of each liquid type during operation. Results revealed strong correlations between FTIR indicators (e.g., oxidation and hydroxyl group loss) and dielectric parameters (permittivity and loss factor), confirming the effectiveness of this combined diagnostic approach. Post-aging breakdown analysis showed that natural esters, particularly NE1215, offered superior preservation of insulation integrity compared to mineral oil. Differences between the two esters also highlight the role of chemical composition in insulation performance. This study reinforces the potential of natural esters as viable, eco-friendly alternatives in thermally and electrically stressed applications.

1. Introduction

Power transformers are at the heart of modern electrical networks, enabling efficient voltage regulation to ensure electricity can be delivered safely and reliably [1]. The durability and performance of these crucial assets rely heavily on the condition of their insulation systems, which traditionally consist of mineral oil and kraft paper [2]. Over time, these materials are subjected to thermal, electrical, and mechanical stress, all of which contribute to gradual aging and eventual degradation of both their insulating and mechanical properties [3]. Among these materials, kraft paper is particularly sensitive to thermal aging. As it degrades, the cellulose structure weakens, leading to reduced mechanical integrity and insulation performance [4,5]. Mineral oil, used both as a dielectric fluid and an impregnation medium, also suffers from aging; it oxidizes over time and produces acidic compounds and sludge, which further accelerate paper degradation and reduce overall system efficiency [6,7,8]. In response to these challenges, growing attention has been directed toward natural esters (NEs) as an alternative insulating liquid. These biodegradable oils provide several advantages over traditional mineral oils, including better thermal stability, higher fire safety due to higher flash point temperature, and improved protection of the paper insulation [9,10,11,12]. Studies show that Kraft paper immersed in natural esters retains a higher degree of polymerization after aging, which suggests better long-term stability [13,14]. Another key benefit of natural esters is their ability to slow down the formation of harmful aging by-products like furans from cellulose insulation, helping to limit acidification and preserve the dielectric properties of the overall insulation system [14,15]. However, despite their promising profile, the behavior of these fluids under electrical stress during thermal aging, particularly breakdown conditions, is still not fully understood. Some reports have noted increases in moisture content and acidity over time, which could affect long-term performance [16,17]. Furthermore, the connection between aging and electrical performance under combined thermal and electrical stress has not been fully explored. Electrical breakdowns, in particular, can drastically change the material properties, often leading to irreversible damage. These events involve localized heating and the formation of discharge paths in the paper, with severe consequences for both dielectric and mechanical performance, especially in mineral oil-impregnated systems [15,16]. Natural esters, with their better thermal behavior, might offer a more robust alternative [17,18,19,20]. Moreover, while FTIR and FDS are often used separately to assess the condition of insulating materials, their combined use remains rare. A multimodal FTIR-FDS approach could enable earlier and more reliable detection of aging phenomena. It is important to note that natural esters are not a homogeneous group, their performance depends on their chemical composition, source (e.g., rapeseed, soybean), and degree of saturation [21,22,23]. Few studies have systematically compared multiple esters under the same thermal aging and electrical stress conditions. Therefore, this study investigates the stability of oil-impregnated paper using both mineral oil and natural esters under simultaneous thermal and electrical stress. Specifically, it presents a detailed comparative analysis of three insulation systems: one based on mineral oil and two based on natural esters (NE1204 and NE1215). The study is structured around the following three main objectives: (i) to evaluate the potential of a multimodal diagnostic approach combining FTIR and FDS for the early detection of insulation degradation mechanisms; (ii) to analyze the impact of electrical breakdown on the chemical and dielectric properties of impregnated paper; and (iii) to compare the two commercial natural esters in order to characterize differences in their chemical and dielectric stability under combined thermal and electrical stress. The findings aim to enhance understanding of aging mechanisms, support the selection of appropriate insulating fluids, and inform the development of predictive diagnostic strategies.

2. Materials and Methods

2.1. Selection and Design of Specimens

The insulating paper analyzed in this study is a kraft paper made from sulfate wood pulp, composed of approximately 90% cellulose polymers, 6–7% hemicellulose, and 3–4% lignin [24]. Circular samples (40 mm in diameter and 0.25 mm thick) of Weidmann Kraft paper were prepared and dried in a vacuum oven (Fisher Scientific) at 105 °C for 48 h. After drying, the samples were impregnated with three types of insulating liquids: a conventional mineral oil (Polaris GX, used as the reference mineral oil in this study) and two natural esters (Midel eN, the commercial name for the ester-based insulating liquids used in this study), one derived from rapeseeds (NE1204) and the other from soybeans (NE1215), both supplied by M&I Materials. Prior to impregnation, the oils were degassed and dried at a controlled temperature of 60 to 70 °C, reducing moisture levels to below 10 ppm in the oils and below 0.5% in the paper.

2.2. Aging Setup

Once impregnated, the prepared samples were placed in sealed metallic vessels and subjected to accelerated aging according to ASTM D1934 [25]. The vessels were kept in a convection oven at 120 °C for 10, 20, and 30 days, respectively. To simulate conditions within power transformers, key aging accelerators like metallic catalysts (particularly copper) were included. Copper conductors were introduced into the vessels to better replicate the actual degradation mechanisms in service-aged transformers, promoting oxidation and hydrolysis reactions that accelerate the deterioration of both oils and papers. In this study, the nomenclature S0 to S3 is used to denote the aging condition of each sample: S0 refers to the fresh (unaged) oil-impregnated sample, while S1, S2, and S3 correspond to samples aged for 10, 20, and 30 days, respectively.

2.3. Characterization

To evaluate the dielectric strength of the oil–paper insulation, a test was conducted in accordance with ASTM D149-20 [26], adapted to the specific characteristics of the materials under investigation. Each sample was placed between two electrodes in a test cell filled with insulation oil to replicate realistic transformer operating conditions. An alternating voltage was then applied and gradually increased at a controlled rate until the dielectric breakdown occurred. Each breakdown test was repeated six times per oil type and aging duration to ensure statistical reliability. After breakdown testing, the dielectric constant and dissipation factor of the impregnated paper and oil were measured using a broadband dielectric spectroscopy system (NOVOCONTROL Alpha-A, fabriqué par Novocontrol Technologies GmbH & Co. KG, basé à Montabaur, en Germany), covering a frequency range from 10³ Hz to 10⁻³ Hz. All measurements were conducted at a constant temperature of 25 °C. Additionally, attenuated total reflectance Fourier-transform infrared spectroscopy (ATR-FTIR) was used to examine chemical and structural changes in the oil–paper insulation system induced by aging and electrical breakdown. Spectra were acquired using an Agilent Cary 630 spectrometer in transmission mode, spanning the 4000–600 cm⁻¹ range, with 16 scans per acquisition and a resolution of 2 cm⁻¹. For semi-quantitative analysis of chemical degradation, the relative intensities of selected FTIR bands were compared by calculating absorbance ratios. The carbonyl band (C=O, ~1710–1740 cm⁻¹) was normalized against the C–O band (~1030–1060 cm⁻¹), which remains relatively stable during aging and serves as an internal reference. Absorbance values were extracted from the spectra using peak maxima, with all measurements performed under consistent baseline correction and acquisition conditions. Figure 1 illustrates the sample preparation procedure and experimental setup.

3. Results and Discussion

3.1. Breakdown Results

Table 1. Breakdown voltage evolution of paper impregnated with different insulating oils (in kV).

	Oil Type
Aging Stage	NE 1204	NE 1215	MO
S0 (fresh oil)	18.5	18.4	16.9
S1 (10 days)	17.6	17.7	16.2
S2 (20 days)	17.2	17.5	15.6
S3 (30 days)	16.5	17	14.1

presents the evolution of the breakdown voltage of kraft paper impregnated with three types of insulating oils, two natural esters (NE 1204 and NE 1215) and one mineral oil (MO), subjected to thermal aging at 120 °C for various durations: S0 (fresh oil), S1 (10 days), S2 (20 days), and S3 (30 days).

In general, a progressive decrease in breakdown voltage as aging time increases was observed, regardless of the insulating oil used (Figure 2). This decline reflects the thermal degradation of the cellulosic paper, leading to a gradual loss of its dielectric strength. However, the rate of degradation varies significantly depending on the impregnation fluid. Paper impregnated with mineral oil shows the greatest reduction in breakdown voltage, dropping from 16.9 kV at S0 to 14.1 kV at S3, a decrease of approximately 16.6%. In comparison, the reduction is considerably less severe with natural esters: NE 1215 decreases from 18.4 kV to 17.0 kV (−7.6%), while NE 1204 drops from 18.5 kV to 16.5 kV (−10.8%). This result confirms the superior thermal stability of natural esters compared to mineral oil. This enhanced performance may be attributed to several factors: the greater ability of natural esters to absorb and retain moisture, their antioxidant properties, and their stronger chemical affinity with cellulose, which collectively help to limit the degradation mechanisms of the paper. Interestingly, even in the initial state (S0), papers impregnated with natural esters exhibit higher breakdown voltages than those with mineral oil, suggesting better initial impregnation and increased compatibility with cellulosic insulation. When comparing the two natural esters tested, NE 1215 demonstrated better performance after 30 days of aging, with a final breakdown voltage of 17.0 kV compared to 16.5 kV for NE 1204. This difference may be attributed to variations in chemical composition or oxidative resistance between the two ester types, although further chemical analysis is needed to confirm this assumption.

3.2. FDS Analysis of Oil/Paper Insulation

Figure 3, Figure 4, Figure 5 and Figure 6 present the results of Frequency Domain Spectroscopy (FDS) measurements. They show the evolution of dielectric properties, including real permittivity (ε′), loss tangent (tan δ), imaginary permittivity (ε″), and frequency-dependent conductivity, as a function of thermal aging and following dielectric breakdown, for paper samples impregnated with the three insulating oils under study. The real part of permittivity (ε′) reflects the material’s ability to store electrical energy, while the imaginary part (ε″) is associated with dielectric losses due to polarization and conduction mechanisms. The dielectric loss tangent (tan δ) quantifies the energy dissipated as heat, and conductivity is influenced by the presence of degradation by-products, such as acids and water molecules, resulting from thermal and electrical stress. These parameters are key indicators for assessing the insulation condition of oil–paper systems.

The FDS measurements presented in Figure 3, Figure 4, Figure 5 and Figure 6 were crucial for monitoring the evolution of the dielectric properties of the oil-impregnated paper across the three aging stages:

➢

Real Permittivity (ε′): A progressive increase in this parameter was observed, especially for mineral oil from stage S1. This rise indicates an enhanced ability of the material to polarize, likely due to an increase in polar groups and residual moisture as a result of aging. In contrast, the variation of ε′ for the natural esters remains lower and more gradual, reflecting better preservation of the cellulose structure.

➢

Dielectric Loss Tangent (tan δ): The evolution of tan δ, particularly in the low-frequency range (1 mHz–10 Hz), emerges as a key indicator of aging. For Polaris GX, a sharp rise begins from S2, suggesting the accumulation of conductive degradation products and moisture. NE 1204 shows a more moderate and gradual increase in the various analyzed parameters, whereas NE 1215 maintains stable tan δ values up to S2, with only a slight rise at S3, indicating very limited aging effects.

➢

Imaginary Permittivity (ε″) and Frequency-Dependent Conductivity: Both ε″ and conductivity increase significantly for mineral oil starting from S2, highlighting greater dielectric losses due to Joule heating and a decline in insulating performance. In contrast, NE 1204 exhibits moderate changes, while NE 1215 consistently shows the lowest values up to S3.

The results indicate that natural esters, particularly NE 1215, significantly mitigate the aging-related deterioration of the dielectric properties of oil-impregnated paper.

3.3. Analysis of FTIR Spectroscopy

The characterization of virgin materials by Fourier-transform infrared spectroscopy (FTIR) establishes a reference baseline for interpreting chemical changes due to thermal aging and electrical stress. The primary materials investigated, insulating oils (natural and mineral) and cellulose paper, exhibit distinct spectral signatures that reflect their molecular composition. Natural ester oil, primarily composed of fatty acid esters, is characterized by prominent absorption bands corresponding to ester functional groups (C=O, C–O–C) and aliphatic chains (

Table 2. Characteristic FTIR absorption bands of virgin natural ester oil [27,28,29].

Wavenumber (cm−1)	Functional Group	Vibration Type
~1740	C=O (ester)	Carbonyl stretching
~2920	Aliphatic CH2	Asymmetric stretching
~2850	Aliphatic CH2	Symmetric stretching
~1465	CH2/CH3	Bending (scissoring)
~1230–1170	C–O–C (ester linkage)	Stretching
~3400	O–H (alcohols, acids, water)	Broad stretching band (hydrogen bonding)

). In contrast, mineral oil, derived from petroleum refining, consists mainly of paraffinic, naphthenic, and aromatic hydrocarbons and lacks ester-related features (

Table 3. Characteristic FTIR absorption bands of virgin mineral oil [30,31].

Wavenumber (cm−1)	Functional Group	Vibration Type
2921	C-H (alkanes)	Asymmetric stretching
2853	C-H (alkanes)	Symmetric stretching
1458	C-H (methylene)	Bending (scissoring)
1376	C-H (methyl)	Symmetric bending
722	C-C (chaînes longues)	Rocking of methylene

). Cellulosic paper shows characteristic bands associated with O–H and C–H bonds, as well as the carbohydrate backbone of cellulose (

Table 4. FTIR bands characteristic of cellulose paper [31,32].

Wavenumber (cm−1)	Functional Group	Vibration Type
~3330	O–H (cellulose)	Broad stretching (H-bonded hydroxyl)
~2900	Aliphatic C–H	Stretching
~1640	Adsorbed H2O	Stretching of bound water
~1430	CH2 (amorphous cellulose)	Bending
~1375	C–H/CH3 (carbohydrate backbone)	Bending (symmetric)
~1160–1030	C–O/C–O–C (carbohydrate structure)	Stretching (cellulose polysaccharide network)

). These reference spectra serve as a basis for identifying degradation phenomena, such as hydrolysis, oxidation, impregnation, or molecular structure loss, in aged or electrically stressed materials.

Figure 7, Figure 8 and Figure 9 present the ATR-FTIR spectra of the fresh and aged samples before and after dielectric breakdown.

The examination of the evolution of FTIR spectra of the oils after thermal aging reveals chemical transformations that are specific to each insulating fluid:

➢

For NE 1204 and NE 1215, a progressive attenuation of the ester carbonyl band (C=O ~1740 cm⁻¹) occurs alongside the emergence or enhancement of a carboxylic acid band (~1710 cm⁻¹), particularly from stage S2 onward. These changes reflect partial hydrolysis of triglycerides [33]. The quantitative analysis revealed that the relative intensity of the 1710 cm⁻¹ band increases more significantly in NE 1204, reaching a ratio of 0.82 at S3, compared to 0.60 for NE 1215. This finding suggests a more pronounced acid formation in NE 1204.

➢

For mineral oil, characteristic oxidation bands in the 1650–1750 cm⁻¹ range appear as early as S1, with increasing intensity through to S3. It was observed that a peak at 1705 cm⁻¹, associated with ketones and weak acids, becomes predominant, with a ratio of 1.21 at S3, indicating advanced and early oxidation processes that are not solely moisture-induced.

Figure 10 and Figure 11 present the ATR-FTIR spectra of paper samples impregnated with the different oils, both in the fresh state and after aging, before and after dielectric breakdown.

The FTIR spectra of impregnated paper uncovered several structural changes induced by thermal aging (

Table 5. Summary comparison of aging and breakdown effects on the oil–paper systems.

Oil Type	Δε′ (Relative Increase)	ΔTan δ (Low Frequency)	O–H Band (~3300 cm−1)	C=O Band (~1705–1740 cm−1)	Interpretation
Mineral oil	High	High	−38%	+154%	Strong degradation (dielectric and chemical)
NE 1204	Moderate	Moderate	−22%	+92%	Partial degradation, moderate oxidation
NE 1215	Low	Slight	−13%	+49%	Good stability, limited aging effects

):

➢

Progressive reduction of the O–H band (~3300 cm⁻¹) indicates the breaking of hydrogen bonds, reflecting a loss of molecular cohesion and hydrophilicity. We found this decrease is most pronounced for Polaris GX (−38% between S0 and S3), compared to −22% for NE 1204 and only −13% for NE 1215.

➢

Decline of C–O bands (~1030–1060 cm⁻¹), associated with the glucosidic structure of cellulose, signals a degradation of the polysaccharide backbone. The measurements show the intensity loss reaches −40% at S3 for Polaris GX, indicating accelerated depolymerization. In contrast, we observed the reductions are moderate for NE 1204 (−27%) and minimal for NE 1215 (−14%), demonstrating better preservation of cellulose chains.

➢

An increase in carbonyl (C=O) bands on the paper is most evident with Polaris GX, as confirmed by the C=O/C–O intensity ratio, which rises from 0.28 at S0 to 0.71 at S3 (+154%). Notably, this increase is lower for NE 1204 (+92%) and NE 1215 (+49%), confirming that paper oxidation is significantly more advanced when impregnated with mineral oil.

3.4. Comparative Spectral and Multimodal Analysis

The wealth of results obtained from the different analytical techniques, breakdown voltage testing, frequency domain spectroscopy (FDS), and Fourier-transform infrared spectroscopy (FTIR), show strong consistency, revealing a marked interdependence between the chemical, electrical, and dielectric aspects of degradation.

➢

Breakdown behavior: it was observed that the mineral oil Polaris GX leads to the greatest reduction in breakdown voltage, indicating accelerated degradation of the solid insulation. This loss in dielectric strength aligns closely with the early chemical deterioration of the paper that we observed via FTIR.

➢

Dielectric properties (FDS): Through the analyses, the ranking of the oils based on key dielectric parameters (loss tangent, conductivity, and permittivity) follows the order of decreasing performance: NE 1215 > NE 1204 > Polaris GX. This hierarchy directly reflects the extent of aging experienced by each oil–paper system.

➢

Chemical degradation (FTIR): The FTIR spectra corroborate these trends, with increasing levels of oxidation and hydrolysis observed from NE 1215 (most stable) to Polaris GX (most degraded), both in the oil and the impregnated paper.

This convergence of findings highlights a fundamental principle: higher chemical stability of the oil–paper system (i.e., low oxidation and hydrolysis) corresponds to better preservation of dielectric properties and breakdown strength. In other words, chemical degradation serves as a reliable predictor of functional deterioration. Among the tested liquids, natural ester NE 1215 stands out as the best-performing solution, benefiting from a favorable synergy between chemical stability and dielectric properties. Based on the findings, it is believed these results position NE 1215 as a highly promising alternative to mineral oils for extending the service life of transformers under thermal stress (

Table 6. Summary comparison of aging and breakdown effects on the oil–paper systems.

Oil	Breakdown Voltage Drop (S0→S3)	tan δ Evolution (FDS)	Paper Oxidation (ATR-FTIR)	Oil Chemical State (ATR-FTIR)	Aging Rate	Overall Performance
Polaris GX	−16.6%	Significant rise from S2 to S3	Marked from S1, accentuated at S3	Significant oxidation from S1	Fast	Low
NE 1204	−10.8%	Gradual increase	Moderate, mainly from S2	Mild acid formation	Moderate	Good
NE 1215	−7.6%	Slight increase only at S3	Low, visible only at S3	Stable, minimal evolution	Slow	Very good

).

The combined analysis of FDS and FTIR results revealed significant correlations between chemical and dielectric parameters. In particular, the increase in loss factor (tan δ) and permittivity is associated with an increase in carbonyl groups (C=O) and a reduction in hydroxyl groups (O–H) in the FTIR spectra. This observation confirms that chemical degradation, especially oil oxidation and cellulose hydrolysis, can predict the loss of dielectric performance.

Furthermore, post-breakdown analyses highlighted an aggravation of degradation, particularly in the case of mineral oil. The drop in O–H bands and the marked increase in C=O bands indicate irreversible degradation of the paper. Natural esters, particularly NE 1215, demonstrated a greater ability to preserve cellulose structure even after breakdown (electrical stress).

Finally, the comparison between NE 1204 and NE 1215 revealed notable differences. NE 1215 exhibited higher chemical stability (lower acid formation, less oxidation) and more stable dielectric behavior, making it a more robust option for demanding applications.

4. Conclusions

This study provides a comprehensive comparison of the combined thermal and electrical stresses behavior of three oil-impregnated paper insulation systems: a conventional mineral oil and two natural esters (NE 1204 and NE 1215). Through a series of dielectric strength tests, FTIR spectroscopy, and frequency-domain dielectric spectroscopy (FDS), this work highlights the complex interdependencies between chemical degradation and dielectric performance.

The first major contribution of this work is the validation of a multimodal diagnostic approach combining FTIR and FDS to monitor early degradation. Strong correlations were observed between chemical oxidation markers, such as the increase in carbonyl (C=O) bands and decrease in hydroxyl (O–H) groups, and key dielectric indicators, including tan δ and permittivity. This finding supports the feasibility of nondestructive, predictive monitoring techniques for assessing insulation health.

Secondly, the study addresses the phenomenon of electrical failures and examines their consequences—a topic that, although previously explored, warrants renewed attention in the context of alternative insulating materials. The results clearly show that electrical breakdown exacerbates both chemical and dielectric deterioration, especially in mineral oil-based systems. The structural damage and oxidation observed in cellulose paper were far more severe after breakdown in mineral oil than in natural ester systems.

Thirdly, the comparative analysis between NE 1204 and NE 1215 reveals that not all natural esters offer the same level of protection and performance. NE 1215 exhibited superior resistance to oxidation, better retention of dielectric properties, and more effective preservation of cellulose integrity, both during thermal aging and post-breakdown. This suggests that the specific formulation and chemical composition of the ester significantly affect long-term insulation reliability.

From an application perspective, the result of this study reveals the advantages of using NE 1215 as a high-performance, biodegradable, and less hazardous insulating fluid. Moreover, the combined use of FTIR and FDS opens new avenues for developing monitoring systems capable of detecting early-stage degradation.

To reinforce and validate these findings, future studies should include degree of polymerization (DPv) assessments, moisture quantification, and investigations under actual transformer operating conditions. These steps will help translate laboratory insights into robust industrial practices for insulation condition assessment and fluid selection in power transformers.