Degradation Mechanisms of Cellulose-Based Transformer Insulation: The Role of Dissolved Gases and Macromolecular Characterisation

Abstract

The ageing of cellulose paper-based transformer insulation is a critical factor influencing the reliability and lifespan of power transformers, as insulating paper is not easily replaced or repaired. Therefore, this review explores the degradation mechanisms of insulating paper, emphasising the roles of dissolved gases, chemical markers, and macromolecular characterisation in assessing paper deterioration. Likewise, the impact of moisture and thermal stress on the breakdown of cellulose fibres are discussed, especially acid hydrolysis, which serves as the main degradation mechanism in cellulose insulating paper. Advanced diagnostic techniques for insulation condition monitoring, such as molecular simulations, glass transition temperature analysis, and DP estimation models, are highlighted. Furthermore, special attention is given to natural esters as alternative insulating liquids, demonstrating their ability to slow cellulose ageing through moisture absorption, hydrogen bonding stabilisation, and transesterification reactions. This paper also evaluates key chemical markers, including 2FAL and methanol, for estimating paper degradation. A comprehensive understanding of these mechanisms and diagnostic approaches can enhance predictive maintenance strategies and improve transformer longevity.

1. Introduction

Paper is a crucial material with significant relevance across various sectors, including information storage, energy transmission, and packaging, as well as tissue and hygiene products. Its economic significance stems from cellulose, a thermodynamically stable biopolymer, which provides exceptional optical and mechanical properties. Cellulose, a natural polysaccharide polymer with the chemical formula C₆H₁₀O₅, forms the primary composition of the transformer insulating paper. Each cellulose molecule consists of multiple β-D-glucopyranosyl units linked by 1,4-β glycosidic bonds [1]. Within each molecule, three free hydroxyl (OH) groups are present on each β-D-glucopyranose unit. When the hydrogen atoms in these hydroxyl groups are nearby (less than approximately 0.28 nm) to adjacent electronegative oxygen atoms, hydrogen bonds are easily established. These hydrogen bonds contribute to the tightly packed structure of cellulose molecular chains, enhancing the structural order. However, as ageing progresses, the hydrogen bonds in cellulose degrade significantly, weakening intermolecular forces and reducing the number of molecular chains, which ultimately leads to fibre surface damage. Furthermore, thermal ageing has a severe impact on the insulating paper fibres, resulting in significant cell wall deterioration. The fibres become thinner, undergo displacement and deformation, and experience structural damage as the outer layers of both the primary and secondary walls break and detach [2]. This degradation leads to the formation of fibre cavities, causing splitting, peeling, and gradual deterioration of the material [3]. With prolonged ageing, the energy required to counteract thermal cracking in liquid-impregnated insulating paper decreases, leading to a reduction in activation energy. Consequently, the internal structural changes in the fibre accelerate, causing fibre branches to break and increasing the amorphous regions of degraded cellulose. This process intensifies the rate and extent of destruction in the crystalline regions of the cellulose, further compromising its integrity [4,5].

Insulating paper primarily consists of cellulose, which is composed of hundreds to thousands of glycosidic bonds. The average number of glycosidic bonds in a cellulose chain, which quantifies the number of repeating structural units within a polymer chain, is referred to as the degree of polymerisation (DP), and it is a well-established parameter used to indicate the ageing condition of insulating paper [6]. However, in practical engineering applications, determining the DP requires powering off the system and suspending the core, making it an offline measurement that cannot be performed on site. It can be determined by the ratio of the polymer molecular weight to the molecular weight of a single repeating unit, expressed in (1) [7].

$$\mathrm{D}\mathrm{P}={\displaystyle \frac{{\mathrm{M}}_{\mathrm{w}}}{{\mathrm{M}}_{\mathrm{o}}}}$$

(1)

The DP can be assessed using the viscosity method, as specified in IEC 60450 and ASTM D4243. This technique involves dissolving the paper insulation sample in cupriethylenediamine (Cuen) and measuring the viscosity of the resulting solution [8]. The specific viscosity (${\upsilon }_s$) is then determined using (2). Once the specific viscosity is obtained, the intrinsic viscosity (υ) is computed using (3).

$${\upsilon }_s={\displaystyle \frac{\mathrm{p}\mathrm{a}\mathrm{p}\mathrm{e}\mathrm{r} \mathrm{s}\mathrm{o}\mathrm{l}\mathrm{u}\mathrm{t}\mathrm{i}\mathrm{o}\mathrm{n} \mathrm{v}\mathrm{i}\mathrm{s}\mathrm{c}\mathrm{o}\mathrm{s}\mathrm{i}\mathrm{t}\mathrm{y}-\mathrm{s}\mathrm{o}\mathrm{l}\mathrm{v}\mathrm{e}\mathrm{n}\mathrm{t} \mathrm{v}\mathrm{i}\mathrm{s}\mathrm{c}\mathrm{o}\mathrm{s}\mathrm{i}\mathrm{t}\mathrm{y}}{\mathrm{s}\mathrm{o}\mathrm{l}\mathrm{v}\mathrm{e}\mathrm{n}\mathrm{t} \mathrm{v}\mathrm{i}\mathrm{s}\mathrm{c}\mathrm{o}\mathrm{s}\mathrm{i}\mathrm{t}\mathrm{y}}}$$

(2)

$${\upsilon }_s=\left[\upsilon \right]·c·{10}^{k·\left[\upsilon \right]·c}$$

(3)

where k is the Martin constant, which has a value of 0.14 for Kraft paper, and c represents the concentration of chemicals in the solution. Finally, the mean DP value ${\overline{DP}}_{\upsilon }$ is computed using the Mark–Houwink equation given in (4), where K and α are the Mark–Houwink constants, with α = 1 and K = 0.007 [9,10].

$$\left[\upsilon \right]=K·{\overline{DP}}_{\upsilon }^{\alpha }$$

(4)

Kraft insulating paper, commonly used for transformer insulation, consists of approximately 90% cellulose, 6–7% hemicellulose, and 3–4% lignin. Due to its low lignin content, Kraft paper is predominantly hydrophilic, allowing for strong bonding between hydrophilic compounds. As a result, it exhibits high mechanical strength and a high DP, making it an ideal insulating material for transformers [1]. A new transformer paper typically has a DP of 1200, while a DP of around 200 indicates the end of its lifespan. The relative rate of depolymerisation (RRD) is determined from the measured DP using (5) [11].

$$\mathrm{R}\mathrm{R}\mathrm{D}={\displaystyle \frac{1200-\mathrm{D}\mathrm{P}}{\mathrm{D}\mathrm{P}}}$$

(5)

The condition of insulating paper insulation is typically assessed through DP and tensile strength measurements. The tensile strength of insulation paper is determined by two key factors, which are the strength of the bonds between the fibres and the strength of the individual fibres themselves. Tensile strength depends on both fibre integrity and the inter-fibre bonding force, while the DP is influenced solely by fibre length. The primary impact of ageing on insulation paper is the alteration of its molecular structure. Over time, the glucose chains within the fibres break down, leading to a reduction in mechanical strength [12,13]. During transformer operation, the insulating paper undergoes degradation due to heat, and electrical stress, as illustrated in Figure 1 [14,15]. This degradation process involves a series of cracking reactions that lead to the removal of hydroxyl groups, the breakdown of glycosidic bonds, and a reduction in hydrogen bond content within the cellulose structure. As a result, the hydrogen bond content in cellulose serves as a key indicator of its ageing state [16]. Furthermore, thermal ageing leads to the deterioration of the paper surface quality. Initially smooth, the microfibril structure of the paper becomes rough and coarse over time as a result of this ageing process. The microfibrils that make up the paper weaken, thin, and eventually disappear with prolonged exposure to heat. In addition, the formation of microcracks and wrinkles further signifies the degradation of the paper caused by accelerated thermal ageing. As a result, the paper becomes more prone to tearing and breaking. In general, the primary degradation mechanisms in cellulose paper occur in three forms, which include pyrolysis, hydrolysis, and oxidation [17].

In pyrolysis, chemical degradation of the paper insulation occurs due to heat in the absence of moisture and an oxidising agent. The byproducts of pyrolysis include water, carbonyl compounds, solid carbon residues, hydroperoxides, carbon monoxide, and 1,6-anhydro β-D-glucopyranose (levoglucosan) [13]. In addition, 1,6-anhydro β-D-glucopyranose can lead to the formation of various acids, such as levulinic acid, acetic acid, pyruvic acid, acetone, methanol, hydrocarbons, and furanic compounds. The average maximum hotspot temperature typically ranges from 95 °C to 110 °C, with the average winding temperature increase being between 55 °C and 65 °C. In cases of overload or cooling system failure, the hotspot temperature inside the transformer may exceed 130 °C, causing the cellulose around the hotspot to undergo pyrolytic degradation [13,18]. Above 130 °C, the predominant degradation process occurs and the direct breakdown of cellulose due to pyrolysis becomes more pronounced at temperatures exceeding 200 °C. However, below this threshold, pyrolysis primarily accelerates other degradation mechanisms, resembling the natural ageing of cellulose but at an increased rate. At the microscopic level, cellulose exhibits lower thermal stability in its C-O bonds compared to the C-H bonds found in insulating liquid. Elevated temperatures can break these C-O bonds, leading to a decline in the DP and a progressive reduction in mechanical strength. Due to the low thermal conductivity of insulating paper, heat tends to accumulate, further promoting degradation. This heat buildup facilitates the cleavage of chemical bonds, generating byproducts such as aldehydes, carboxyl groups, and carbon dioxide. On a macroscopic scale, the pyrolysis rate is influenced by the combined effects of oxygen, moisture, acid concentration, and temperature [19,20]. In the case of hydrolysis, the paper insulation steadily absorbs moisture from the insulating liquid. During operation, moisture transfers between the insulating liquid and the paper, with the paper typically containing higher moisture levels than the insulating liquid under stable conditions. As the moisture content increases, dielectric loss rises, and dielectric strength decreases, which is mainly a result of moisture and the acids that play a crucial role in the hydrolytic degradation of cellulose insulation [20]. Hydrolysis is driven by the presence of hydrogen ions (H⁺) released from the dissociation of acids in water. In the hydrolysis process, acid and water interact, breaking the glycosidic bonds and resulting in the formation of glucose molecules. The accumulation of glucose molecules accelerates the hydrolysis, eventually leading to the formation of furan compounds, like 5-hydroxymethyl-2-furaldehyde (5HMF) [18]. Hydrolysis is the primary degradation mechanism within the temperature range of 70 °C to 130 °C. Due to its hygroscopic nature, cellulose readily absorbs moisture, allowing water molecules to accumulate within its structure as cellulose pyrolysis can generate water as a byproduct. The interaction between water and cellulose drives the hydrolysis process, which is a significant factor in the ageing of insulating paper. During hydrolysis, the 1,4-β-glycosidic bonds connecting glucose units are cleaved, leading to the formation of shorter cellulose chains. These short-chain molecules undergo further hydrolysis, resulting in the production of carboxylic acids and additional water. The cleavage of each glycosidic bond consumes one water molecule and subsequently releases three, yielding a net gain of two water molecules per bond breakage [19]. Also, moisture content significantly influences the rate of paper degradation. Dried Kraft paper samples tend to degrade at a slower rate compared to those with higher initial moisture content. This is because the hydrolysis reaction, which requires water, is less active in dried samples. Consequently, controlling the moisture content in transformer insulation can help mitigate ageing and prolong operational life.

Table 1. Comparison of drying techniques for transformer insulation.

Techniques	Explanation	Strength	Drawback
Indirect drying	- Uses absorbent materials to remove water from transformer liquid, which indirectly dries the solid insulation	- Continuous moisture control - Cost-effective preventive maintenance	- Primarily affects liquid rather than solid insulation
Vapour phase drying	- Uses solvent vapour for heating and condensation, with vacuum pumps enhancing the moisture level	- Efficient moisture removal - Suitable for large transformers	- Complex process as it requires controlled conditions
Hot air drying	- Applies hot air at 100 °C to 110 °C, which is typically used in convection ovens for transformer drying	- Simple and widely employed - Effective for small transformers	- Uneven heat distribution - Limited to small units
Heat and vacuum treatment	- Uses heat and vacuum to lower the boiling point of water, allowing for moisture removal at safe temperatures	- Highly efficient - Prevents insulation degradation and is suitable for all transformers	- Requires specialised chambers and vacuum pumps

presents the techniques employed for drying solid insulation in transformers [7]. Furthermore, moisture can penetrate insulation through three main mechanisms, which are residual moisture from the manufacturing process, moisture absorption from the atmosphere, and the degradation of cellulose caused by ageing. The oxidation process of cellulose insulation is a gradual reaction that results in carbon dioxide and water as byproducts. This process is accelerated by reactive oxygen, as the extent of oxidation depends on the oxidising agents and the pH of the surrounding environment. During transformer operation, the interaction between water and oxygen forms hydrogen peroxide in the presence of copper and iron. The oxidation degradation of cellulose insulation, which predominantly occurs at temperatures below 75 °C, is initiated by primary and secondary hydroxyl groups, leading to the production of aldehydes and acidic complexes as primary byproducts. Also, if the breather fails, the insulating liquid may come into direct contact with air, causing an increase in dissolved oxygen levels in the insulating liquid, which can range from 2000 ppm to 3000 ppm [18]. Cellulose is particularly vulnerable to oxidative degradation, as its OH groups readily convert into carbonyl and carboxyl groups, leading to secondary chain-scission reactions. However, during the ageing of transformer insulating liquid and cellulose, free oxygen atoms are not generated within the transformer. Researchers suggest that transition metal ions, such as Cu⁺/Cu²⁺ and Fe²⁺/Fe³⁺, facilitate the reaction between oxygen and water, resulting in the formation of hydrogen peroxide. Furthermore, hydroxyl radicals produced from hydrogen peroxide decomposition are believed to accelerate the oxidation process [19]. The primary type of cellulose cleavage is illustrated in Figure 2 [21] and Figure 3 shows the diagram illustrating the degradation pathways of paper and their observable effects at various scales [22].

Therefore, considering the various techniques and complexity of the cellulose paper ageing process, this review analyses the degradation mechanisms of transformer insulating paper, focusing on the main mechanism of cellulose degradation, which is acid hydrolysis. In addition, it explores the impact of dissolved gases, chemical markers, and insulating liquids on cellulose ageing. Key predictive models for estimating the DP and advanced diagnostic techniques for insulation assessment are discussed. We anticipate that a better understanding and advancements in insulation ageing mitigation techniques will significantly enhance transformer reliability in the coming years.

2. Cellulose Pulp: Composition and Properties

Cellulose with a chemical structure as seen in Figure 4 is a naturally occurring polymer that is widely distributed across the globe, with an estimated annual production ranging from 10¹⁰ to 10¹¹ tons [13,23]. However, only a small portion, approximately 6 million tons, is utilised in industries, such as paper, textiles, chemicals, and construction. Cellulose is made up of repeating β-(1,4)-linked D-glucopyranosyl anhydroglucose units arranged in a ⁴C₁-chain configuration. Each monomer is rotated by 180° relative to the adjacent unit, forming a crystalline structure called elementary fibrils. These fibrils aggregate to form microfibrils, which further combine to create macro fibrils or cellulosic fibres. The bonding within cellulose, both intramolecular and intermolecular, significantly affects its properties, such as solubility and chirality. Native cellulose is composed of both crystalline and amorphous regions, with its degree of crystallinity varying between 30% and 95%, depending on the source and extraction method. The DP ranges from 10,000 to 15,000 units. Cellulose can be sourced from various materials, including plants, animals, and bacteria. In lignocellulosic biomass, cellulose exists in fibril form, embedded within lignin and hemicellulose [24,25]. Cellulose is degradable, exhibits chirality, and is a hydrophilic compound due to its molecular configuration, making it insoluble in water and many organic solvents. Furthermore, cellulose is odourless; cost-effective; and possesses high crystallinity, remarkable biocompatibility, low density, and significant mechanical strength. The composition and properties of cellulose can differ depending on its source [26,27]. Standardised methods for extracting cellulose from these materials involve processes such as pulping and chemical pretreatments to separate lignin and hemicellulose, resulting in pure cellulose. Lignin, a phenolic compound, forms part of the lignocellulosic biomass and can constitute up to 20% of the material depending on the source. Hemicellulose consists of three primary subgroups: xylan, mannan, and xyloglucan. Pulping, often under neutral conditions, is the most commonly used pretreatment for extracting cellulose [24,25]. So, electrical-grade papers are produced from unbleached softwood pulp, which is derived using the Kraft cooking process. This process involves breaking down wood chips under heated, pressurised, and alkaline conditions to obtain wood pulp. Several factors, including raw material composition, temperature, chemical concentration, and cooking duration, influence the quality of the resulting pulp. During the Kraft cooking process, hemicellulose and lignin are partially removed to separate the fibres from the wood matrix, which alters the mechanical properties of the final paper product. The removal of lignin enhances fibre bonding, resulting in stronger paper. However, this process also causes some degradation and a loss of cellulose, which weakens the fibres. As a result, Kraft cooking requires a balance between achieving sufficient delignification and preserving cellulose yield and its DP. The extent of lignin removal is commonly assessed using the kappa number as expressed in (6), which measures the residual lignin content in the pulp and is also influenced by the presence of hexenuronic acid.

$$\mathrm{k}\mathrm{a}\mathrm{p}\mathrm{p}\mathrm{a}\mathrm{n}\mathrm{u}\mathrm{m}\mathrm{b}\mathrm{e}\mathrm{r}\approx {\displaystyle \frac{\mathrm{w}\mathrm{t}\mathrm{\%} \mathrm{l}\mathrm{i}\mathrm{g}\mathrm{n}\mathrm{i}\mathrm{n}}{0.15}}$$

(6)

where the ratio depends on both the pulp type and processing conditions. Kraft pulp used for electrical-grade paper typically has a kappa number between 20 and 28, corresponding to a lignin content of approximately 3–4.2 wt% [28]. When wood pulp undergoes drying, it experiences hornification, which is a structural change that prevents the fibre from fully regaining its original porous structure after rewetting. Consequently, dried pulp has weaker mechanical properties compared to never-dried pulp. Mechanical beating can enhance fibre fibrillation and increase the presence of fines, thereby improving paper strength after pressing. However, it also reduces the pulp’s ability to release water during both the forming and drying stages of paper manufacturing. Lignin removal significantly affects the morphology, density, and chemical composition of the final paper product. Since lignin contains aromatic structures, its polarisation properties are different from those of cellulose and hemicellulose. This suggests that lignin content may influence the performance of the material [28]. The summary of different types of insulating paper is presented in

Table 2. Summary of different types of insulating paper.

Insulating Paper	Manufacturing Process	Composition	Strength	Limitation	Thermal Class	Ref.
Kraft paper	Kraft process/sulphate process	Cellulose (primary component), hemicellulose, lignin (partially removed), and residual sulphates from Kraft process	- Cost-effective and widely available - Slightly alkaline, aiding insulation longevity	- More vulnerable to thermal, mechanical, and electrical stress	Class A (105 °C)	[3,13,29,30]
Crepe paper	Kraft process with an extra creeping step	Cellulose, hemicellulose, and modified with crimping (mechanical process) to enhance extensibility	- Higher stretch (elongation) capacity to wrap around copper wiring without tearing - Used to insulate irregular shapes and connections - Provides reliable pairing of insulation layers during winding	- More expensive than standard Kraft paper - Less durable under prolonged mechanical stress	Class B (130 °C)	[3,29,31,32,33]
Thermally upgraded paper	Kraft process with additional nitrous compound	Cellulose, hemicellulose (partially replaced), and stabilising nitrogen compounds (dicyandiamide, urea, melanine)	- Better thermal performance than standard Kraft paper - Slower ageing rate (up to 2.5 times slower than Kraft paper) - Higher resistance to hydrolysis and oxidation - Partially neutralise the attack of the acids against the cellulose and reacts in contact with water	- More expensive than standard Kraft paper - Reduced mechanical strength due to hydroxyl group substitution - Some chemical processes s(cyanoethylation modification) involve toxic nitrile organic compounds	Class A (105 °C)	[3,29,34,35,36]
Diamond-dotted paper	Kraft process with an epoxy layer	Kraft paper-based, thermosetting resin (diamond-pattern coating)	- The epoxy layer provides internal strengthening of the coil for better adhesion to the conductor - High mechanical strength due to large bonding surface	- More expensive than standard Kraft paper - Increased manufacturing complexity	Class A (105 °C)	[3,31]
Synthetic paper (Aramid/Nomex)	A blend of wood pulp, synthetic fibres, and a binder	Aramid fibres (polymeric aromatic polyamides), aramid pulp, and no cellulose content	- Mostly polyaramid-based papers with better thermal and mechanical performance for traction transformers - Higher thermal rating than thermally upgraded Kraft (TUK) - Resistance to moisture, acids, and abrasion - Higher bubble initiation temperature than cellulose-based papers	- Low moisture adsorption but easy to hydrolyse at high temperatures - More expensive than other alternatives - Less flexible compared to crepe paper - The aramid portion of the paper cannot be dissolved by most solvents	Nomex 910 Class B (130 °C) Nomex 410 Class C (220 °C)	[3,29,31,36,37,38]

.

The authors in [28] study to fully understand how the composition of lignin impacts the dielectric properties of cellulose paper as they considered the effects of changes in both density and morphology due to delignification. Three pulps having 107, 75, and 27 kappa numbers were employed for this analysis. The outcome of the study shows that lignin of low content should be considered for low dielectric dissipation factor. Therefore, they concluded that insulating liquid influences the charge transport behaviour of insulating paper in liquid, but the chemistry of the pulp as related to its kappa number likewise influences the dielectric loss, polarisation, and charge transport.

3. Marker to Capture DP

While transformer insulating liquid can be periodically replaced, the insulating paper is not easily replaced, making it the primary factor that dictates the transformer lifespan. However, several markers, as shown in

Table 3. Comparison among the chemical markers from cellulose-based paper.

Markers	Strength	Limitation	Ref.
Carbon oxides	- Widely used in transformer diagnostics - Excellent indicator for overheating fault as it is being generated through the thermal decomposition of insulating paper - A standardised online technique with dissolved gas analysis (DGA) - C-O bond within the glucose ring undergoes cleavage to form a carbonyl group and release CO2 molecules - Dehydrogenation of the secondary alcohol hydroxyl group on the glucose ring forms a carbonyl group and then close bonds weaken to release CO molecules	- Not specific to cellulose degradation as they can also originate from insulating liquid oxidation - Low sensitivity to early stages of paper degradation - Different rates of diffusion of CO from paper to liquid at higher temperature - Concentration influenced by leakage from the atmosphere in a free-breathing transformer - Can be generated by partial discharges in the transformer	[19,40,41,42,43]
2-furfural (2FAL)	- Specific to cellulose degradation - Concentration is sensitive to cellulose depolymerisation in mineral and ester liquid - Popularly used in the industry	- Unreliable for TUK due to the influence of alkaline inhibitors, which interfere with the acid hydrolysis reaction responsible for 2FAL formation - Can originate from any pyranose-based compound as its production may result from hemicellulose degradation and not just cellulose breakdown - Only accurate in the later stages of degradation, when acid hydrolysis has reduced the cellulose DP to approximately 400 or lower - Not stable above 110 °C - Concentration affected by the presence of water, acids, and oxygen - There is an effect of humidity on the rate of production	[3,40,42,44,45]
Ethanol (EtOH or CH3OH)	- Can indicate localised faults such as hotspots for abnormal cellulose degradation - Serves as a hot-spot chemical marker for cellulose insulation - Has a higher concentration than MeOH at a higher temperature - Generation not influenced by environmental parameters - Helps to differentiate between abnormal and normal ageing	- Not a general indicator of paper ageing - Generated more from insulating liquid ageing than paper - Related to abnormal paper ageing that occurs at very high temperatures since it is generated from levoglucosan, which is the byproduct of pyrolysis - Particularity of the EtOH behaviour appears at temperatures of over 250 °C	[39,40,45]
Methanol (MeOH or CH3CH2OH)	- Effective for both Kraft and TUK paper - Capable to detect early paper ageing - More stable among paper degradation byproducts compared to acetone, butanol, and ethanol - Linked to the rupture of 1,4-β-glycosidic bonds of cellulose - The time required to recover MeOH after processing the liquid of a specific transformer is half the duration needed for the recovery of 2FAL	- High volatility may lead to loss over time - May undergo esterification with LMA from paper and liquid, which will affect its accuracy - Requires multiple experimental validations, including stability, partitioning, and origin studies - Requires extensive validation with DP and tensile strength - The rate of methanol formation stabilises and no longer increases significantly when the DP is below 400 - Shows a partial sensitivity to the cellulose ageing in ester liquid	[3,40,44,45,46]
Dispersion staining colour	- Identifies ageing stages through colour transition	- Depends on a different colour ratio ageing marker - Subject to variations in lighting and imaging conditions - May not have a universal calibration for all insulation types	[47]
Refractive index	- Correlates with the degradation of cellulose fibres - Can be studied using molecular simulation software	- Can be influenced by bulk density and polarisation changes - Requires specialised equipment and simulation tools - May not be directly measurable in in-service transformers	[47]
Total sugar	- Sugar is one of the principal decomposition products of cellulose - Can serve as an early ageing marker than 2FAL	- Different types of sugar are dissolved in the insulating liquid, so further investigation is required to determine the total sugar concentration - Unreliable to judge the paper ageing by the concentration of one type of sugar - May not be detected and measurable in real transformers, since sugar decomposes rapidly and may not persist long	[48]

have been proposed to assess the condition of the insulating paper insulation system in transformers. Furthermore, among these markers, significant efforts have been made to explore the potential and limitations of furanic compounds and methanol as chemical markers for assessing the ageing of cellulose insulation in power transformers. For furanic compounds, reports from the IEEE and CIGRÉ have recently addressed this subject. Their key finding suggests that among the five furanic compounds identified by Burton et al. in 1984, 2-furfural (2FAL) offers the most valuable insight into paper degradation. However, factors such as the amount of solid insulating material and transformer design complicate the interpretation of 2FAL levels and the establishment of standard threshold values. Furthermore, methanol, proposed in 2007 as a marker for insulating paper ageing, was detected during a laboratory ageing experiment, where its presence was linked to the scission of glycosidic bonds in the cellulose polymer structure [39].

Furthermore, various physicochemical factors, including temperature, humidity, and oxygen concentration, can impact the 2FAL concentration in transformer liquid. Previously, drawing firm conclusions about the reliability of 2FAL as a diagnostic tool was challenging due to the absence of comparable markers. Recently, methanol (MeOH) has emerged as a new marker for transformer insulation diagnostics. This marker is advantageous because it is produced from all types of cellulose-based papers, such as standard Kraft and thermally upgraded papers, even at low temperatures, unlike 2FAL. Moreover, MeOH generation is directly linked to the rupture of the cellulose 1,4-β-glycosidic bond, a relationship not demonstrated with any other marker. As a result, its presence may serve as an indicator of the residual life of insulating paper. More recently, ethanol (EtOH), also studied by the same research group examining MeOH, has exhibited distinct behaviour in certain transformer liquid analyses when compared to lab-based ageing experiments. Figure 5 gives the key characteristics of an ideal chemical marker [3].

4. Characterisation of Paper Ageing

4.1. Molecular Dynamics Simulations

Cellulose molecules exhibit polarity, leading to strong interactions between their molecular chains. Due to the abundance of free OH groups in cellulose, extensive hydrogen bonding occurs. These intramolecular hydrogen bonds restrict the rotation of glycosidic bonds, resulting in a rigid structure that enhances the mechanical properties of cellulose insulation paper. In molecular dynamics (MD) simulations of cellulose, factors such as model structure, DP, force-field selection, simulation step size, dynamic ensemble, and boundary conditions play a crucial role [49]. According to a study in [50], the crystalline regions of cellulose exhibit greater stability, while the amorphous regions are more heterogeneous, with degradation initiating in the amorphous phase. Furthermore, the dielectric properties and thermal stability of cellulose molecules are primarily influenced by the amorphous region, making amorphous models the preferred choice for most cellulose simulations.

4.2. Glass Transition Temperature

The glass transition temperature (Tg) is a crucial indicator that characterises the properties of polymer materials. It represents a dynamic transformation process where the material undergoes significant changes. Below Tg, the material remains relatively stable, but once the temperature surpasses this threshold, molecular mobility, degrees of freedom, strength, and hardness exhibit noticeable alterations. Also, the mechanical, thermal, electromagnetic, and optical properties of the material are affected. One of the most widely used and reliable techniques for determining Tg is the specific volume–temperature curve method. This approach involves mapping the temperature corresponding to a specific volume, as obtained from NPT molecular dynamics simulations. The glass transition temperature is identified at the inflexion point where the specific volume changes with the temperature, determined through linear fitting before and after this transition [51].

5. Cellulose Degradation

Cellulose, the primary component of insulation paper, is a linear high-molecular-weight compound consisting of β-D-glucosyl units linked by 1,4-glycosidic bonds. Its fundamental repeating unit is cellobiose, and its chemical structure follows the empirical molecular formula (C₆H₁₀O₅)n, where n represents the DP. The composition of cellulose includes carbon, hydrogen, and oxygen, with respective mass fractions of 44.44%, 6.17%, and 49.39%. In insulating paper, cellulose exists in two structural forms, which are crystalline and amorphous regions. The crystalline region features tightly packed and well-ordered molecules, whereas the amorphous region contains irregularly arranged molecules with weaker intra- and intermolecular forces and numerous voids. When a transformer is in operation, thermal ageing of the insulation paper primarily begins in the amorphous region. Also, research has shown that cellulose chains contain weak points in approximately every 500 glucose monomer units, predominantly located in the amorphous regions. These regions are more permeable, allowing water and acids to penetrate and accelerate degradation. In contrast, the crystalline regions are more resistant to hydrolysis [52].

It was reported in [53] that the initial thermal degradation of insulation paper starts in the amorphous regions, leading to molecular chain breakage, surface cavities, increased roughness, and a reduction in the DP. As a result, enthalpy rises, while the increase in relative crystallinity causes a decrease in activation energy. Over time, ageing significantly damages the crystalline regions, reducing relative crystallinity and leading to a decline in enthalpy. Furthermore, the roughness of a transformer insulating material influences the insulating liquid flow electrification during operation. Increased surface roughness enhances surface charge absorption, leading to greater charge accumulation as liquid flow intensifies. Also, higher roughness improves water absorption, which can degrade insulation performance and accelerate ageing. This, in turn, increases the risk of insulation damage under high temperatures.

5.1. Acid Hydrolysis

Acid hydrolysis acts as the primary parameter in cellulose degradation, which occurs in the presence of water and acids. It is initiated when an acid donates a proton to water, yielding hydronium, which then transfers the proton to cellulose and initiates chain splitting. As proposed by the Bronsted–Lowry acid-base theory, the acid can likewise transfer the proton directly to cellulose without involving water [54]. However, as water is more polar and has a greater difference in electronegativity between oxygen and hydrogen than cellulose does between oxygen and carbon, it then serves as a better proton acceptor. Once the proton is transferred, it interacts with oxygen in the cellulose until water facilitates the chain scission. Also, hydronium is damaging as the resulting water molecules can continue to react with the cellulose structure after donating a proton to cellulose. Notably, a water molecule is consumed during chain scission, while the H⁺ ion is regenerated, keeping its concentration unchanged. Therefore, based on this reaction mechanism, the rate of paper degradation is solely attributed to the concentration of both water and H⁺ ions dissociated from acids [55,56]. Figure 6 illustrates the fundamental chemical reaction mechanism underlying the acid hydrolysis of cellulose, as it shows that the rate of degradation of insulating paper is a factor of the moisture content and the number of free H⁺ in the acid molecules [19], and Figure 7 shows the mechanisms through which an acid decomposes cellulose paper. Based on this reaction mechanism, the rate of paper degradation is influenced by the concentration of water and the availability of H⁺ dissociated from the acids [54]. The oxidation of insulating liquid generates two types of acids, which are high-molecular-weight acid (HMA), and low-molecular-weight acid (LMA), as seen and further explained in

Table 4. Comparison between LMW and HMW acids.

Acid	Properties	Type	Structure	Ref.
LMAs	- Produce by insulating liquid oxidation and are highly soluble in insulating liquid - Are hydrophilic and react chemically to give a proton (H+) - Significant influence on the paper degradation rate - Readily interact with cellulose and water due to their ability to form bonds with the hydroxyl (OH) group - Readily dissolve in cellulose paper and, at the same time, damage the crystalline and amorphous parts of the paper - Have a high diffusion rate - Influence water distribution between paper and insulating liquid - Contain a strong polar carboxyl group and a non-polar alkyl group - Are also known as short-chain organic acids	Formic acid		[19,54,55,56,57]
		Levulinic acid
		Acetic acid
HMAs	- Hardly soluble in insulating liquid - Not aggressive and are hydrophobic causing them to have a lower affinity for cellulose - Little participation in insulating paper degradation - Can deposit on radiator pipes, reducing heat transfer efficiency - Have little or no influence on water distribution between paper and insulating liquid - Have a low diffusion rate - Contain a strong polar carboxyl group and a non-polar alkyl group - Are also known as long-chain organic acids	Stearic acid (CH3(CH2)16COOH)		[19,55,56,57]
		- Naphthenic acid (R(C5H8)(CH2)nCOOH

.

In [55], the authors studied the influence of LMA on the degradation of mineral oil-impregnated Kraft cellulose paper considering the different LMAs, which are formic, levulinic, and acetic carboxylic acid. Naphthenic-based mineral oil was used as well as samples with acid concentrations of 0.2 mg KOH/g and 0.40 mg KOH/g, which contain the three acids either as an individual or in a different combination of these acids. It was observed from their result that the insulating paper absorbed formic acid more readily than the acetic and levulinic acids, which results in a higher ageing rate for insulating paper exposed to formic acid-blended insulating liquid. This study can be further explored, as it was observed that the purity levels of the different carboxylic acids used varied significantly. Acetic acid exhibited the highest purity at 0.997, followed by levulinic acid at 0.98, while formic acid had the lowest purity at 0.88. This discrepancy in purity levels may have influenced the outcomes and should be critically examined in future research.

Acid Detection

An accurate analysis of the concentration of LMA along with the water content in the insulating liquid and paper has been proposed over time to be highly valuable for assessing the ageing conditions and evaluating the impact of maintenance action [56]. Although it was impossible to monitor the concentration of the acid in insulating paper, this acid value in paper can be estimated from its value in insulating liquid. The total acid number (TAN), having a standard unit of measurement as mg KOH/g, measures the acid concentration in a non-aqueous solution. The measurement process involves determining the amount of potassium hydroxide (KOH) needed to neutralise the acid present in one gram of a liquid sample [55]. The acid value in transformer insulating liquid arises from the degradation of both the insulating liquid and cellulose. However, the acids and water produced can interact, accelerating the hydrolysis of the insulating paper and further contributing to acid formation. As illustrated in Figure 8, the acid hydrolysis of cellulose generates free glucose molecules, which subsequently break down into hydroxyl methyl furfural, levulinic acid, formic acid, and other acidic compounds. The accumulation of acids in the insulating liquid leads to an increase in its conductivity while diminishing its insulating properties. In addition, it can impact the insulating liquid heat dissipation capability and contribute to the ageing of paper insulation materials. Since both insulating liquid and paper undergo ageing simultaneously, their degradation processes are likely interrelated. This suggests a possible correlation between the acid value of the liquid and the DP of the insulating paper [58]. Figure 9 illustrates the recommended acid values for insulating liquid reclamation, as proposed by IEC 60422 [55]. The primary methods for detecting the concentration of insulating liquid-dissolved acetic acid include indicator-potentiometric titration and water-soluble acid determination. The indicator-potentiometric titration method requires pretreatment steps such as extraction and heating in a water bath, making it complex, time-consuming, and subjective. On the other hand, the water-soluble acid determination method suffers from poor reproducibility due to the sensitivity of pH values to both experimental water and indicators. Furthermore, neither method can distinguish between different acid types or their strengths [56].

In [56], Raman spectroscopy is used to directly analyse dissolved acetic acid in insulating liquid, which helps to eliminate the complexity of the extraction process. The techniques focus on identifying the characteristic Raman-shift spectral lines for acetic acid, optimising detection parameters, and validating the techniques with experimental results. The study successfully identified a characteristic Raman shift for acetic acid dissolved in an insulating liquid. The techniques demonstrated a detection limit of 0.68 mg/mL and achieved an accuracy of 91.66% compared to the conventional indicator-potentiometric titration techniques and water-soluble acid determination.

5.2. Number of Ruptured Bonds Within Cellulose

The number of hydrogen bonds in cellulose paper is influenced not only by macroscopic properties like tensile strength but also by its microscopic structure, which plays a crucial role in characterising the ageing of cellulose paper. The normalised scission (NS) model, as seen in (7), is the average number of glycosidic chain scissions per cellulose polymer and is a molecular-level characteristic that can be used to investigate the kinetics aspect of paper degradation.

$$\mathrm{N}\mathrm{S}={\displaystyle \frac{{DP}_{\left(v0\right)}}{{DP}_{\left(vt\right)}}}-1$$

(7)

where ${DP}_{\left(v0\right)}$ and ${DP}_{\left(vt\right)}$ are the initial DP and the DP at the ageing time of t [39]

In [39], the variation in methanol and 2FAL in insulating liquid was analysed against the normalised scission to compare the two chemical markers during the early ageing period. It was observed that methanol gives a linear variation with NS and the quantity of methanol is higher than 2FAL, confirming that methanol is suitable for the early degradation state of cellulose paper. The study in [59] illustrated that the application of an electric field leads to a reduction in the total number of hydrogen bonds within the cellulose. It can be inferred that the presence of an electric field causes an expansion of the cellulose chains while diminishing hydrogen bonding, ultimately increasing the spacing between cellulose chains and resulting in a less compact structure. On a macroscopic scale, this phenomenon manifests as a decrease in the DP and a deterioration in the mechanical properties of the insulation paper over time. Furthermore, exposure to an electric field reduces the cellulose average molecular potential energy while increasing the mean square displacement (MSD) of the cellulose chains. This also contributes to a decline in hydrogen bonds and accelerates the ageing of the insulation paper, with this effect being more prominent at lower temperatures. One possible explanation is that higher temperatures weaken the influence of the electric field.

5.3. Yield Behaviour of Cellulose Paper

The cellulose paper used is primarily Class A sulfate pulp paper, composed of α-cellulose and hemicelluloses, where 70% and 30% of the α-cellulose exist in a crystalline and an amorphous state respectively. This makes the cellulose insulation paper possess a two-phase system with cross-linked crystalline and amorphous regions. Since the crystalline region remains stable even at high temperatures, the thermal and mechanical strength response of the cellulose paper are mainly determined by the amorphous region. It has been reported that the impact of an electric field on the crystalline structure is minimal due to the strong binding effect of the microscopic lattice system on the cellulose chain. Consequently, the amorphous region is considered the primary focus for studying the effects of an electric field. Therefore, authors in [60] investigate the microscopic behaviour of cellulose insulation paper under an electric field, where single-chain and multi-chain cellulose models were developed using molecular dynamics simulations. They reported that studies on the effects of electric fields on insulation paper have relied on macroscopic tests, leaving the microscopic mechanisms largely unexplored. So, their study tends to bridge that gap by modeling the yield behaviour of cellulose under a strong electric field. Their results show that both single- and multi-chain cellulose structures exhibited yield behaviour when subjected to a strong electric field. The single-chain cellulose displayed irregular yield patterns, forming a V-like shape, with both ends aligning in the same direction as the electric field while the central region oriented oppositely. In contrast, the multi-chain cellulose aligned consistently with the electric field direction. The stress induced by the electric field ultimately led to the breakdown of the cellulose chains, which highlights the impact of electrical stress on insulation paper degradation.

Also, ref. [61] explores the relationship between the DP, fracture degree, and mechanical strength of cellulose insulation paper. Fracture degree is introduced as a new concept to better understand the mechanical degradation of cellulose. Nine cellulose crystal models with varying fracture degrees were constructed, and their mechanical properties and hydrogen bond numbers were analysed using molecular dynamics (MD) simulations. The results indicate that as insulation paper ages and cellulose chains break, the elastic constant significantly affects Young’s modulus, and a decrease in DP and an increase in fracture degree lead to a stepwise reduction in Young’s modulus. Furthermore, as the fracture degree increases, the average number of hydrogen bonds decreases, directly influencing mechanical strength since hydrogen bonding is a key factor in maintaining cellulose integrity. MD simulations under transformer operating conditions revealed a strong positive correlation between mechanical parameters, all of which decline in a ladder-like manner with a decreasing DP and an increasing fracture degree. The study confirms that axial strength reduction in cellulose chains is the most critical factor affecting the modulus and that mechanical strength is primarily influenced by the direction of the cellulose chain. Also, the motion of cellulose chains within the lattice constraints has little effect on mechanical strength, further validating that the hydrogen bond network plays a crucial role in maintaining cellulose stability.

5.4. Temperature Condition

Cellulose degradation occurs in three distinct temperature stages, as seen in Figure 10. In the initial stage (25–150 °C), the insulation paper transitions from physically adsorbing water to desorbing it. In the second stage (150–240 °C), dehydration affects certain glucose residues within the cellulose structure. This process generates water, along with small amounts of carbon dioxide and carbon monoxide, while also forming carbonyl and carboxyl groups. The presence of oxygen significantly impacts these reactions. In the final stage (240–400 °C), glycosidic bonds break, which leads to the depolymerisation of cellulose. This results in the formation of new products and low-molecular-weight volatile compounds with fibre carbonisation [62].

5.5. Effect of Hydrogen Sulphide

The presence of hydrogen sulfide in transformers is primarily linked to the ageing process of both crude oil and insulating liquid–paper insulation. Sulfur compounds in crude oil can be categorised into three groups based on their reactivity: active sulfides, neutral sulfides, and inactive sulfides. Among these, active sulfides are particularly corrosive to transformer insulation systems, with hydrogen sulfide itself classified as an active sulfide. In addition, hydrogen sulfide molecules can be generated within the insulating liquid–paper insulation system through pyrolysis, as well as the thermal degradation of sulfur-containing compounds, such as mercaptans and thioethers. In some cases, the breakdown of disulfides may also lead to the release of free sulfur or hydrogen sulphide [51,63].

In [51], the glass transition temperature of cellulose was recorded at 462 K, 441 K, 410 K, 395 K, and 379 K when the hydrogen sulfide content was 0%, 2%, 4%, 6%, and 8%, respectively. These results indicate that increasing hydrogen sulfide content leads to a decrease in the glass transition temperature. When the hydrogen sulfide concentration reached 8%, the glass transition temperature dropped significantly. The primary cause of this reduction is the formation of numerous hydrogen bonds between hydrogen sulfide and cellulose, which interferes with and alters the cellulose’s original hydrogen bonding structure. Furthermore, the presence of hydrogen sulfide molecules diminishes the thermal stability of cellulose. As the concentration of hydrogen sulfide increases, the interactions between cellulose chains become weaker, leading to a linear rise in the diffusion coefficient of hydrogen sulfide. This disrupts the existing hydrogen bond network within the cellulose structure. Consequently, the mobility of cellulose chains is enhanced, further reducing its thermal stability. In addition, a high hydrogen sulfide concentration weakens the van der Waals forces between cellulose molecules, intensifying chain movement and further compromising the thermal stability of cellulose. The study in [63] investigated the corrosion mechanism of sulfides in insulating liquid and their impact on liquid–paper insulation deterioration. Specifically, it examined the effects of dibenzyl disulphide (DBDS), hexadecanethiol, and their combination on insulating paper. Experimental samples were thermally aged to analyse changes in tensile strength and the DP of the insulating paper. Also, reactive molecular dynamic simulations were conducted to study the microscopic mechanism of sulfide-induced degradation in cellulose materials. The results showed that the combination of DBDS and hexadecanethiol caused greater degradation in the tensile strength and degree of polymerisation of the insulation paper compared to using a single sulfide. Simulations confirmed that sulfides significantly weaken cellulose molecules, with hexadecanethiol having a stronger effect than DBDS when used individually. The molecular energy calculations revealed that the degradation of cellulose mechanical properties was mainly due to a reduction in hydrogen bonds and van der Waals interactions. This structural weakening explains why the combination of DBDS and hexadecanethiol had the most severe impact on the insulation paper.

5.6. Cellulose Degradation in Insulating Liquid

In [64], the enhancement in the thermal ageing properties of cellulose aged in natural methyl ester is a factor of two modes of protection, which are the steric hindrance of the cellulose hydrolytic reaction and water scavenging. According to the forming, thermal stress can cause the natural ester to interfere with the cellulose hydrolysis process. This interference occurs due to the reaction of the natural ester with the OH groups on the glucose ring and at chain termination. The proposed mechanism involved transesterification, driven by temperature conditions during accelerated ageing. Esterification of the reactive OH sites on cellulose with bulky ester groups stabilises the molecules, thereby extending their thermal life. According to the latter, water scavenging occurs in natural ester through two mechanisms. First, at elevated temperatures, natural esters can undergo hydrolysis, consuming water from the cellulose and thereby decreasing the potential moisture-related damage. Second, natural esters have a significantly higher water-holding capacity than mineral insulating liquids. While the water saturation limit at room temperature is approximately 1050 mg/kg for natural ester, it is only about 60 mg/kg for mineral insulating liquid. As a result of this difference, natural esters can absorb more water, enhancing their paper-drying effect, but tend to increase the acid number. As discussed earlier, the principal type of acid in natural ester is high molecular acids, which does not significantly influence the ageing of insulating paper. Therefore, considering these two mechanisms, it was reported that thermally upgraded Kraft paper (TUK) ages considerably slower in natural ester insulating liquid than in mineral insulating liquid under the same thermal condition [65]. Also, in [66], it was explained that the lower degradation observed in paper immersed in natural ester at higher temperatures could be a result of the deposition of a gel-like substance on insulating paper, which is produced by the degradation of natural ester that shields the insulating paper from further deterioration.

In [67], a comparative analysis was conducted on the TUK impregnated with natural ester, synthetic ester, and mineral insulating liquid. The study found that TUK ages at the slowest rates in synthetic ester, followed by natural ester, and then mineral. Also, in an oxygen-free system, excluding other water-generating transformer materials, it was observed that the lifespan of TUK impregnated with natural ester at 110 °C is at least 1.6 times longer than that of TUK impregnated with mineral oil. This outcome was attributed to the stronger hydrogen bonds between the molecules of synthetic ester or natural ester and TUK, which tends to stabilise TUK from decomposition. Moreover, the transesterification reaction between the reactive OH group on the cellulose molecules and the fatty acids in natural ester, facilitated by hydrolysis, further inhibits cellulose degradation. The water-scavenging capability and hydrolytic protection provided by natural ester also contribute to the extending lifespan of the TUK. Furthermore, the authors concluded that the primary mechanism of TUK protection by the synthetic ester and natural ester is through the water and acid translocation from the TUK to the esters, which results in a drier and cleaner TUK, effectively preventing the auto-catalysed hydrolysis of TUK and slowing down degradation. The study in [68] investigated the thermal ageing of Kraft paper in natural ester and mineral oil under sealed conditions at 105 °C for 365 days by comparing the DP, tensile strength, breakdown voltage, dissipation factor, and sludge formation. It was reported that the natural ester significantly slows paper degradation, retains higher tensile strength, maintains better breakdown voltage, produces minimal sludge, and preserves insulating liquid clarity. The suppression of cellulose Kraft paper degradation in natural ester compared to mineral oil (MO) is attributed to three key factors. First, the natural ester-saturated chemical structure provides high oxidative stability, preventing the formation of large amounts of water and acids that would otherwise accelerate paper decomposition, as seen in mineral oil. Second, hydrogen bonding between natural ester and cellulose stabilises the paper by lowering its energy state, thereby increasing the activation energy required for oxidation compared to mineral oil. Lastly, polar substances in natural ester scavenge water and acids, preventing their accumulation and further protecting the Kraft paper from degradation, unlike mineral oil, which allows for the build-up of small polar molecules that catalyse cellulose breakdown. The authors in [69] found that the ageing rate of natural ester-immersed paper was slower than that of mineral oil-immersed paper, as evidenced by higher DP values. In addition, initial moisture content had a significant impact on mineral oil-immersed samples but was less influential in natural ester-immersed samples. A notable decrease in water content was observed in natural-immersed paper, especially at higher temperatures and moisture levels, whereas mineral oil-immersed paper maintained a stable water content. The key factors differentiating natural ester-immersed paper ageing from mineral oil-immersed paper ageing include moisture equilibrium between insulating liquid and paper, ester hydrolysis, and paper transesterification.

The study in [70] investigated the thermal properties of cellulose and aramid paper impregnated with different liquid insulations, which are mineral oil, synthetic ester, and natural ester, in power transformers to identify the best combination of paper and liquid insulation for effective heat transfer. This was achieved by measuring the thermal expansion, specific heat, density, thermal conductivity, kinematic viscosity, and heat transfer coefficient of each insulating liquid. The study revealed that cellulose paper exhibited higher thermal conductivity than aramid paper across all liquid types. Among the liquid insulations, natural ester provided the highest thermal conductivity, followed by synthetic ester and mineral oil. Consequently, the most effective combination for heat transfer in paper insulation was cellulose paper impregnated with natural ester. In terms of liquid insulation, mineral oil demonstrated the highest heat transfer coefficient, outperforming both esters. Although natural ester had better thermal conductivity than mineral oil, its significantly higher viscosity negatively affected its heat transfer efficiency. Among the esters, natural ester performed better at lower temperatures less than 60 °C, whereas synthetic ester was slightly more effective at higher temperatures. Regarding overall transformer cooling efficiency, cellulose paper impregnated with natural ester emerged as the best choice for paper insulation due to its superior thermal conductivity. However, mineral oil proved to be the most effective liquid for heat transfer, offering a better balance of properties despite its lower thermal conductivity. The optimal combination for transformer insulation, considering both paper and liquid performance, was cellulose paper with mineral oil. In conclusion, cellulose paper impregnated with natural ester is the most suitable option for paper insulation due to its excellent thermal conductivity. Meanwhile, mineral oil remains the most effective liquid insulation for achieving optimal heat transfer. A trade-off exists between the two: natural ester enhances paper insulation performance, whereas mineral oil ensures better overall cooling efficiency in transformer systems. This study provides valuable insights for selecting insulation materials to enhance transformer performance and longevity.

The study in [71] evaluated the compatibility of paper insulation with natural ester by examining its dielectric properties and thermal stability after impregnation. The findings revealed that paper impregnated with natural ester exhibited a higher relative permittivity and a lower dissipation factor, both of which contribute to enhanced dielectric performance. In terms of thermal stability, the analysis indicated that both natural ester and natural ester-impregnated paper had a higher initial decomposition temperature compared to their mineral oil counterparts. This suggests that natural ester-based insulation can endure greater thermal stress. Regarding moisture content, the study found that paper impregnated with natural ester retained less moisture than mineral oil-impregnated paper. This is attributed to the higher water solubility of natural ester, which absorbs moisture from the cellulose, keeping it dry. Since water accelerates cellulose degradation, its reduced presence in natural ester-impregnated paper slows down the degradation process. In contrast, mineral oil retains more moisture within the paper, leading to faster deterioration. The DP of aged insulation was also assessed, revealing that natural ester-impregnated paper experienced a smaller reduction in the DP compared to mineral oil-impregnated paper, even under thermal stress. This suggests lower cellulose degradation, which can be attributed to reduced moisture content and the superior thermal conductivity of natural ester, allowing for more efficient heat dissipation and slowing down degradation. The study in [72] investigated the effects of accelerated thermal ageing on insulation paper immersed in different insulating liquids, including mineral oil and ester-based liquids. Also, a microcosmic mechanism analysis was conducted to understand the interactions between insulation materials and the surrounding insulating liquid medium. The results indicate that the DP values of insulation cellulose immersed in soybean liquid remain higher throughout the ageing process compared to those in palm insulating liquid and mineral oil. The degradation rate of insulation cellulose in mineral oil is found to be 2.59 times greater than in soybean liquid and 1.03 times higher than in palm liquid. Molecular simulations reveal that mineral oil does not attract hydronium ions but rather repels them. In contrast, hydrogen bonding interactions between esters and hydronium ions are significantly more pronounced compared to those in mineral oil. Moreover, the hydrogen bonding between cellulose and hydronium ions in natural esters is weaker than that observed in mineral oil. With an increasing concentration of hydronium ions in the liquid–paper composite system, the ability of natural esters to restrict these ions eventually reaches saturation, potentially diminishing their protective effect on cellulose insulation. However, soybean liquid demonstrates the highest capacity to tightly restrict hydronium ions, offering superior protection to cellulose insulation paper. Based on the findings, soybean insulating liquid exhibits the best performance in preserving cellulose insulation integrity during thermal ageing. The ability of natural esters to regulate hydronium ion interactions plays a crucial role in their protective properties. Furthermore, mineral oil exhibits low polarity, which prevents it from attracting hydronium ions. As a result, hydronium ions migrate to the cellulose and form hydrogen bonds, compromising the structural integrity of the cellulose chains. In contrast, palm insulating liquid has a higher polarity, allowing it to attract hydronium ions. However, its strong polarity leads to the formation of hydrogen bonds with cellulose, destabilising the cellulose chain structure. Soybean insulating liquid, with its optimal polarity, can effectively retain hydronium ions, leading to a greater number of hydrogen bonds within the cellulose, thus enhancing the stability of the cellulose chains. Overall, soybean insulating liquid provides the best performance in extending the lifespan of insulation paper due to its strong attraction to hydronium ions and its ability to form the most hydrogen bonds in cellulose, aligning with the observed DP values. Furthermore, according to [73], natural esters containing a higher number of carbon-carbon double bonds are generally more effective in preventing ageing. This suggests that, beyond acid catalysis and moisture migration, carbon-carbon double bonds could play a significant role in slowing down the ageing process by inhibiting the oxidative degradation of insulation paper.

Table 5. Summary of ageing studies on insulating paper impregnated in insulating liquid.

Ref	Insulating Paper	Insulating Liquid	Ageing Temperature	Test	Finding	Explanation
[29]	- Hybrid paper (aramid/cellulose) - TUK	- Corn-based natural ester - Mineral oil	165 °C and 185 °C	- Tensile strength - DP	- Hybrid paper has better performance than TUK in both liquids - Slower degradation observed in paper in a natural ester	- No transesterification reaction in both papers impregnated in insulating liquid
[38]	- Hybrid paper (cellulose/Aramid) - TUK	- Soya bean-based natural ester - Mineral oil	165 °C, 175 °C, 185 °C, and 195 °C	- Tensile strength - DP - Acidity - Moisture content - Viscosity - Power loss	- Final ageing duration of TUK in ester liquid is 5.9 times longer than that of TUK in mineral oil and 4.6 longer than hybrid paper in mineral oil; this indicates that TUK in ester liquid exhibited significantly greater resistance to ageing compared to the other two combinations	- Hybrid paper, which has the least effect on insulating liquid degradation, shows that the degradation of insulating liquid depends on the type of insulating paper
[74]	- Crepe paper	- Natural ester - Naphthenic mineral oil	150 °C for seven months	- Tensile strength - DP - Viscosity - Breakdown voltage	- Rapid deterioration was observed during early ageing but more pronounced in mineral oil (53%) than natural ester (27%) - Decrease in DP is less pronounced when using natural than mineral oil	- There is greater protection of cellulose when in natural ester as it consumes moisture from the cellulose due to hydrolysis
[75]	- TUK - Diamond-dotted paper - Mixture of wood pulp and cotton paper (Grade 3)	- Sunflower liquid (Natural ester)	150 °C for 5232 h	- Dielectric loss Permittivity - Resistivity - Moisture content - DP	- TUK and diamond-dotted paper exhibited similar levels of degradation - Grade 3 paper experienced the most severe deterioration - Insulating liquid showed greater degradation when aged with Grade 3 paper compared to the other two papers	- Grade 3 paper performs worst because of the high acidity detected in its samples despite the fact that natural ester generates high molecular acid, which has no significant influence on cellulose degradation
[76]	- Kraft paper - TUK	- Natural ester - Mineral oil	150, °C, 160, °C, 170, °C, and 180 °C	- DP - Tensile strength - Furan and MEOH - Acidity - Interfacial tension - Dielectric loss	- Furan concentration was observed to be 10 times lower in ester liquid than in mineral oil - Concentration of methanol was reported to be much lower in ester liquid. - Reported that natural ester insulating liquid extends the lifespan of Kraft paper and TUK as compared to mineral oil	- More furans in ester liquid may be due to their consumption in the reaction with nitrogen compounds in the TUK - There is a stronger bonding of ester groups to cellulose OH groups, which limits the dissolution of furans in ester liquid - More MeOH produced may be a result of the reaction of methanol with acids
[33]	- Crepe paper	- Aromatic mineral oil - Synthetic ester (Midel 7131)	90 °C, 110 °C, and 130 °C for 2000 h	- DP	- Report reveals that, within the temperature of 90 °C to 110 °C, the depolymerisation of cellulose occurred most extensively in the mineral oil - Incorporating synthetic ester into mineral oil at a concentration of 10% and 30% by volume enhances the thermal stability of both the insulating liquid and the paper impregnated in the blends	- Result obtained may be due to hydrolysis reactions in synthetic esters, which lead to the formation of long-chain fatty acids that do not have a negative influence on insulating paper - Ester groups block the cellulose OH sites, creating a protective barrier against the negative effect of polar water molecules
[77]	- Hybrid paper (aramid/cellulose) - TUK	- Mineral oil - Natural ester	165 °C, 175 °C, and 185 °C	- Tensile strength - Young’s modulus - Elongation	- Natural ester enhances the mechanical characteristics of the TUK and hybrid paper at all aged temperatures - TUK reaches the 25% retained tensile strength target faster than hybrid paper in both mineral oil and natural ester; this suggests that hybrid paper may have a longer lifespan than TUK	- Report may be due to natural ester capacity to offer a protective role for cellulose, which preserves the elasticity of the interlamellar amorphous phase by controlling van der Waals interactions among cellulose chains and promoting selective water absorption
[21]	Cellulose Kraft paper	- Mineral oil - Vegetable insulating liquid	Not stated	- Number of hydrogen, water, and formic acid molecules	- The number of H2, H2O, CO, and formic acid molecules in the insulating paper is diminished when vegetable insulating liquid is present, with a notable decrease compared to when mineral oil is used	- The higher number of hydrogen bonds, binding energy, and electrostatic energy between vegetable insulating liquid and paper, compared to mineral oil, is the reason why there is a stronger interaction between vegetable insulating liquid and paper
[78]	- Kraft paper - TUK	- Mineral oil - FR3 ester liquid	120 °C	- DP - Furan content	- Report indicates that Kraft paper and TUK impregnated in FR3 exhibited higher DP values and slower ageing rate - Substitution of mineral oil with FR3 leads to a notable reduction in the chain scission of both papers - FR3 demonstrates a more significant ageing-retarding effect on TUK than on Kraft paper	- The observation was analysed based on atomic-scale mechanisms, which reveal that FR3 ester liquid effectively restricts the diffusion of water and acid within the insulating liquid–paper insulation system; this inhibition plays a crucial role in slowing down the acid hydrolysis of cellulose
[79]	- Kraft paper - TUK - Aramid paper	- Mineral oil	Between 100 °C and 180 °C	- Bubbling inception temperature	- TUK insulating paper exhibits the highest bubbling inception temperature, followed by Kraft and aramid paper; this observation shows that TUK has the highest resistance to the formation of paper in the insulating liquid compared to the other two insulating papers	- It was reported that the bubbling inception temperature is highly dependent on the water content in the insulating paper and pressure, as these factors determine the threshold at which trapped water within the fibres vapourises and generates bubbles
[80]	- Kraft paper	- Mineral oil - Natural ester	130 °C and 150 °C	- DP - Tensile strength - Water content - AC breakdown voltage - Dielectric loss	- Report shows that natural ester exhibits a slower degradation rate, with only a 27% reduction after 8000 h, in contrast to mineral oil, which experiences a 58% decrease after 3500 h - Natural ester displays superior performance in dielectric loss, AC dielectric strength, and moisture content	- The observation shows that the high water saturation level of natural ester has a drying impact on the insulating paper, which helps to slow its degradation rate as compared to mineral oil with low-water-level saturation

gives the summary of the ageing studies on insulating paper impregnated in insulating liquid and

Table 6. Summary of the mechanism employed to enhance the ageing of cellulose.

Mechanisms	Summary	Ref.
Water absorption	Natural ester absorbs water from the transformer paper due to its high water solubility. This reduces the moisture in the insulating paper, slowing down its ageing process. The water saturation level of the natural ester is significantly higher than that of mineral oil, helping to maintain moisture balance.	[65,66,67,69]
Promotion of hydrogen bond	Natural ester enhances hydrogen bonding between cellulose fibres, strengthening the structure of the insulation paper. Also, it reduces the water content in the paper, preventing water molecules from disrupting the hydrogen bonds and thereby stabilising cellulose by strengthening the C-O and C-C bonds in the cellulose.	[65,66,67]
Reduction in hydrolysis	Cellulose degradation involves hydrolysis, where water competes with hydroxyl groups on the cellulose sugar ring for protons. Since natural ester absorbs more water from the paper, it reduces this competition, thereby slowing down hydrolysis.	[65,66,67,69]
Restriction in pore swelling	Natural ester helps to maintain the ordered structure of cellulose molecules, preventing excessive swelling of fibril pores. This helps preserve the mechanical integrity of the paper insulation as it causes the neighbouring open pores in the cellulose to swell.	[65,66]
Water consumption through hydrolysis	Natural ester undergoes hydrolysis, consuming available water in the insulation system. This reduces the damaging effects of water on cellulose, unlike mineral oil, which does not react with water and allows for moisture accumulation.	[65,66,67]
Esterification of insulating paper	During ageing, long-chain fatty acids produced from natural ester hydrolysis can esterify the cellulose paper, chemically modifying its structure. This reaction enhances the paper stability and provides additional protection against degradation.	[65,66,67,69]

summarises the mechanism employed by natural ester in enhancing the ageing of cellulose in transformer insulation [65].

6. Key Ageing Characteristics: Natural Ester vs. Mineral Oil-Impregnated Paper

6.1. Insulating Liquid–Paper Moisture Equilibrium

In natural ester, water absorption alters the moisture equilibrium between insulating paper and liquid, leading to higher water retention in the liquid at equilibrium. As a result, the moisture content in natural ester reaches a peak at a certain stage of ageing. This effect is particularly evident in papers with high initial moisture content exposed to elevated temperatures. Therefore, natural ester absorbs more water in solution compared to mineral oil [69]. In thermal ageing tests, the moisture equilibrium in insulation paper impregnated with mineral insulating oil differs from that of vegetable or mixed insulating liquids. Due to the high water solubility and hydrolysis process of vegetable insulating liquid, more moisture migrates from the insulation paper into the insulating liquid. This process effectively dries the insulation paper and minimises the impact of moisture on its degradation rate [81].

6.2. Hydrolysis of Esters

The key distinction between the ageing processes of natural- and mineral oil-immersed paper stems from the way natural ester and mineral oil degrade. While mineral oil primarily undergoes oxidation, natural ester exposed to high temperatures predominantly degrades through hydrolysis. The hydrolysis reaction of esters, illustrated in Figure 11, involves the breakdown of a triglyceride using three water molecules, producing glycerol and three long-chain fatty acids. Hydrolysis reactions are influenced by both temperature and moisture content. Experimental data in [69] supported this by showing that as hydrolysis consumes water dissolved in a natural ester, the equilibrium shifts, leading to more water dissolving in the insulating liquid rather than the paper.

6.3. Transesterification Reaction

The hydrolysis reactions in the natural release long-chain fatty acids, which tend to bond with cellulose through a process known as transesterification. This process involves the substitution of the reactive OH groups of glucose molecules with fatty acids as seen in Figure 12. Transesterification of glucose monomers, which has been observed experimentally using SEM and studied theoretically through molecular simulation, slows down the depolymerisation of paper insulation by preventing the cleavage of esterified glucose monomers from the chains. Figure 13 illustrates the cleavage process of a non-esterified glucose chain, which occurs in three stages: First, an OH group reacts with a hydrogen atom of an adjacent carbon group, forming a water molecule. Second, the alcohol is oxidised into a ketone, with the hydrogen from the OH group displaced towards the carbon group. Lastly, the CH₂OH group reacts with the oxygen linking glucose molecules, cleaving the chain. For transesterified glucose molecules, the CH₂OH group is blocked by fatty acids, halting the reaction after the second step and preventing chain cleavage [69].

7. Models for DP Prediction

The degradation of transformer insulation paper is a key indicator of a transformer’s remaining lifespan. Since a direct measurement of the DP is impractical in an operating transformer, various models have been developed to estimate DP based on the concentration of 2FAL in the insulating liquid. These models consider different factors, such as temperature effects, moisture content, paper type, and liquid conditions.

Table 7. Model to capture DP of insulation paper.

Models	Formula	Strength	Limitation	Ref.
Cheim-Dupont	$DP={\left({\displaystyle \frac{2FAL}{\lambda }}\right)}^{{\displaystyle \frac{1}{\Psi d}}}$ λ—shortening expression d—parameter representing the type of paper and on the winding longitudinal temperature gradient	- Considers both hotspot temperature and paper type - Incorporates the effect of temperature gradients inside the transformer	- Does not account for localised degradation - Insulating liquid regeneration or replacement alters 2FAL concentration - Does not directly consider the effect of oxygen and moisture on paper ageing - Typically applicable to mineral oil - Depends on transformer design	[82,83]
Chendong	$DP={\displaystyle \frac{\log \left(2FAL\right)-1.51}{-0.0035}}$ 2FAL is in ppm	- Developed from real transformer data, making it useful for practical diagnostics - More reliable than using only hotspot gradient since furans are direct byproducts of paper ageing - Account for DP between 150 and 1000	- Limited to transformers with normal Kraft paper - Limited to transformers with free-breathing conservators - If insulating liquid has been replaced or regenerated, furan concentration changes, leading to inaccurate DP estimation - Typically applicable to mineral oil	[82,84]
Stebbins	$DP={\displaystyle \frac{\log \left(2FAL\times 0.88\right)-4.51}{-0.0035}}$ 2FAL is in ppb	- Modified Chendong model by using TUK - More reliable than using only hotspot gradient since furans are direct byproducts of paper ageing	- Only applicable to certain transformers depending on the type of insulation paper - Assumes that furans fully represent paper degradation, ignoring other influencing factors - Typically applicable to mineral oil	[82]
De Pablo 1	$DP={\displaystyle \frac{1850}{2FAL+2.3}}$ 2FAL is in ppm	- Accounts for the fact that paper rarely decomposes evenly - Considers the effect of hotspot and thermal gradient - Polymer degradation main chain theory is utilised	- Accounts for DP between 150 and 600 - Typically applicable to mineral oil	[85]
De Pablo 2	$DP={\displaystyle \frac{7100}{8.88+2FAL}}$ 2FAL is in ppm	- It is a simple linear equation - Utilises the theory of cellulose chain scissions - Accounts for the fact that paper rarely decomposes evenly	- Paper is assumed to degrade uniformly - The initial DP of the paper was assumed to be 800, which is lower than the typical value in a real transformer - Typically applicable to mineral oil - Assumes molecular weight of 2FAL is 96, three cellulose chain scission give one 2FAL molecule, and the ratio of insulating liquid to paper is 25	[82,83,84,86]
Pahlavanpour	$DP={\displaystyle \frac{800}{\left(0.186\times 2FAL\right)+1}}$ 2FAL is in ppm	- Accounts for non-uniform paper ageing within transformer winding - Modifies De Pablo model	- Assumes a fixed 20% of inner paper layers degrade twice as fast as the remaining 80% of the paper, which may not reflect real-world variations - Typically applicable to mineral oil - Furan concentration is affected by insulating liquid replacement or reconditioning process	[82,83]
Vaurchex	$DP={\displaystyle \frac{2.6-\log 2FAL}{0.0049}}$ 2FAL is in ppm	- Model was developed based on experimental results - Accurate results for transformers less than 8 years in operation	- Does not utilise data from a real-world transformer - Typically applicable to mineral oil - The experiment was conducted exclusively on Kraft paper	[83,84,85]
Burton	$DP={\displaystyle \frac{2.5-\log 2FAL}{0.005}}$ 2FAL is in ppm	- Model was developed based on experimental results	- The experiment was conducted exclusively on Kraft paper - Typically applicable to mineral oil - Does not utilise data from a real-world transformer	[83,84]
Chaouhui	$DP=405.25-(347.22\times \log 2FAL$) 2FAL is in ppm	- Based on real-world transformer data - Suitable for various transformer conditions	- Limited dataset used for model development - Typically applicable to mineral oil- Does not consider external influencing factors, like moisture	[87]
Myers et al.	$DP=1288.6-(285.7\times \log (2FAL\times 0.88)$) 2FAL is in ppb	- For non-TUK paper - Accurate results for transformers less than 8 years in operation	- The presence of other furan compounds aside from 2FAL indicates unusual paper degradation - Typically applicable to mineral oil	[84,85]
Myers et al.	$DP=1387.5-(373.8\times \log 2FAL$) 2FAL is in ppb	- For TUK paper	- 2FAL has low stability in paper treated with dicyandiamide - Typically applicable to mineral oil	[85]
Serena	$DP={\displaystyle \frac{7100}{8.88+2FAL+1}}$ 2FAL is in ppm	- Validates De Pablo model by suggesting that one 2FAL molecule is generated for every three broken cellulose chains	- Assumes a fixed relationship between furan generation and paper breakdown, which may not always be accurate - Typically applicable to mineral oil	[87]
Li Song	$DP=-121\times \ln 2FAL+458$2FAL is in ppm	- Accurate result for older transformers around 30 years in operation	- Assumes a specific degradation pattern for aged transformers, which may not generalise well - Typically applicable to mineral oil	[85]
Heisler and Banzer	$DP=325\times \left({\displaystyle \frac{19}{13}}-\log 2FAL\right)$	- Account for DP between 100 and 900	- Developed from a limited dataset, restricting broader applicability - Typically applicable to mineral oil	[19]
Dong et al.	$DP=402.47-\left(220.87\log 2FAL\right)$ 2FAL is in ppm	- Accurate result for older transformers around 30 years in operation	- Model may not be suitable for newer transformers - Typically applicable to mineral oil	[85]
Shkolnik et al.	$DP={\displaystyle \frac{\log 2FAL-417}{0.00288}}$	- Model exclusively for TUK	- Model is exclusively for TUK - Typically applicable to mineral oil	[86]
Garcia et al.	$DP={\displaystyle \frac{1}{-0.008}}\log \left({\displaystyle \frac{2FAL}{78.98MP-23.3}}\right)$ 2FAL is in ppm MP—moisture content of the paper in %	- Considers the impact of moisture content in the evolution of 2FAL - Model developed from real transformer data, making it real for practical diagnostics - Accelerated thermal ageing was performed to validate the model	- Not widely used by utility - The type of insulating paper used was not mentioned - Validation was performed using Kraft paper - Typically applicable to mineral oil	[88]
Thomas Leibfried et al.	$DP={\displaystyle \frac{1.4-\log 2FAL}{0.003}}$ 2FAL is in ppm	- Gives accurate results between DP of 200 and 800 - Model developed with real transformer data through post-mortem techniques	- The number of investigated transformers is comparably low for a statistical data evaluation - Typically applicable to mineral oil - More accurate for a grid transformer than a generator set-up transformer	[89]
Jalbert el al	$DP={\displaystyle \frac{1100}{1+1.5542\left(1-e^{-5.4M}\right)+1.01M}}$ M—methanol in ppm	- Model built on a real operating transformer	- Only 6 transformers were considered - Typically applicable to mineral oil	[10,90]
Ghoneim	$DP=1294.4-\left(122.6\times \ln 2FAL\right)$ 2FAL is in ppb	- Provides estimates for DP in aged transformers - Developed from empirical data	- Assumes a fixed relationship between 2FAL and DP, which may not account for real-world variations - Typically applicable to mineral oil	[91]

presents a comparison of several DP estimation models, highlighting their formulas, strengths, and limitations. Each model varies in its assumptions and applicability, making them suitable for different transformer conditions and ageing scenarios.

The discrepancies regarding the use of 2FAL-based models for predicting the DP can be attributed to various factors, such as the following:

• The complexity of degradation and furan generation mechanisms and variability in stress factors (thermal and electrical).
• Differences in the analytical methods (sampling, detection, and calibration standards) that might vary between laboratories or studies.
• Variability in the transformer design, materials, and operational conditions, which complicates the generalisation of 2FAL as a reliable indicator across different studies.
• Disagreements on the reliability and interpretation of the relationship between 2FAL levels and transformer degradation (linear vs. non-linear models depending on factors such as the severity of stress, temperature, and operational conditions).
• The simplification of models that may not account for the full range of factors influencing transformer ageing. Over time, the characteristics of the transformer oil change due to oxidation, moisture absorption, and other chemical processes. Aged oils may have different chemical properties that affect 2FAL concentration, making it harder to draw direct correlations with the transformer’s degradation state. In addition, the presence of additives or impurities in the transformer oil, such as antioxidants or other chemical stabilisers, may interfere with the formation of 2FAL or change its concentration, leading to variation across studies using different types of oil formulations.

To overcome these challenges, future research could focus on developing more robust multi-variable models that incorporate other gas concentrations (e.g., CO, CO₂, and CH₄), temperatures, operational histories, and comprehensive degradation mechanisms, combined with standardised analytical techniques and long-term field validation to improve the reliability of 2FAL-based DP predictions for power transformers.

7.1. Limitations and Factors to Consider for 2FAL

The establishment of relationships between furanic derivatives, other chemical markers, and paper degradation is essential for accurate interpretation. However, the reliability of these models and their diagnostic effectiveness depend on understanding the insulation paper involved. In addition, factors like insulating liquid acidity and moisture content significantly influence the formation kinetics of these markers, making it impossible to develop a dependable model. Furthermore, the kinetic of the marker formation may vary depending on the equipment operating in the nitrogen atmosphere. Also, it has been reported that measured 2FAL concentrations in transformer insulating liquid samples are usually significantly lower than those obtained from laboratory ageing results for the same DP. This discrepancy is said to be a factor of insulating liquid quality and type, which affect the quantity, solubility, and stability of the markers [86,92]. Also, it has been observed that there is a significant difference in the level of 2FAL between the inhibited and uninhibited insulating liquid. As a result, it has been observed that inhibited insulating liquid tends to have much lower 2FAL concentrations relative to the uninhibited insulating liquid. This suggests that inhibitors may influence 2FAL formation and affect its migration from insulating paper to insulating liquid [86]. In addition, the inaccuracy of 2FAL in assessing the TUK condition is attributed to the presence of alkaline inhibitors, which suppress acid hydrolysis and consequently reduce furan formation. Moreover, 2FAL lacks specificity, as it can originate from any pyranose-based compound, including hemicelluloses. It is considered a reliable indicator only at advanced stages of degradation, typically when the cellulose DP has dropped to approximately 500 or lower [3,93]. Therefore, due to these limitations, the transformer’s remaining life may not be reliably predicted based solely on these marker contents. However, estimating the condition of the paper based on changes in furanic compound concentrations in the insulating liquid is then feasible if the sampling is strictly controlled. Figure 14 presents the technical parameters that affect the generation of furan derivatives [83].

7.2. Limitations and Factors to Consider for Methanol

Given the three key limitations associated with 2FAL, particularly as the application of TUK becomes more prevalent in transformers, methanol (MeOH), identified by Jalbert et al. in 2007, has emerged as a promising third-generation marker aimed to address the shortcomings of 2FAL [93]. The strength of MeOH as a degradation marker is summarised in Table 3, which highlights its potential to offer more reliable insights into cellulose ageing.

However, despite its promise, several important considerations must be taken into account when evaluating MeOH effectiveness. These include its reduced stability at high temperatures above 110 °C due to interactions with oxidation byproducts in the insulating liquid oxidation, the possibility of esterification with organic acids, and the variability in measurements during advanced ageing stages, particularly when the DP falls below 200. In addition, challenges include potential losses during sampling due to volatility, delayed detection resulting from partitioning behaviour between the insulating liquid and paper, reduced chemical stability in synthetic ester liquids, and limited real-world validation of existing laboratory-based kinetic models [93,94]. While MeOH represents a significant advancement in the search for more accurate ageing markers, its practical application in field conditions still requires further validation, particularly through long-term studies to ensure reliability and consistency in real transformer environments. After the development of ASTM D8086-20 [95], the next logical steps should involve widespread adoption, validation in field conditions, and integration with other diagnostic methods. A need for international harmonisation and periodic revisions based on user feedback and technological advances is also essential. By ensuring the method becomes a standard practice in the industry, developing complementary standards, and incorporating new technologies, the testing of methanol in insulating liquids may become an important tool in ensuring the safety, reliability, and longevity of electrical systems worldwide.

8. Perspectives

Monitoring the condition of paper insulation in transformers is crucial for ensuring these critical electrical components’ longevity and safe operation. Early-stage deterioration often requires advanced diagnostic tools, which may not always be available or feasible for regular use. Most monitoring methods are indirect and often require specialised equipment and expertise, and the data interpretation can be complex. Continuously monitoring the condition of the paper requires setting up long-term monitoring systems that track various parameters over time. Sophisticated models are needed to account for various influencing factors.

It is recommended that representative samples of the real transformer paper–liquid arrangement be employed in experimental procedures, ensuring that the insulating materials are identical to those in the transformer and that the weight ratio of cellulosic insulation to insulating is preserved. The initial DP of the cellulosic material should correspond to the manufacturer-specified value after drying, and the drying and impregnation process should follow the same technological procedure used during transformer manufacturing. In addition, the insulating liquid used must be free of measurable furan derivatives, and long-term thermal stress should be applied within a temperature range of 130 °C to 140 °C to accelerate investigations. Furthermore, it is suggested that the experiment should be concluded when the DP value decreases to approximately 400. For effective model development, a reliable database should include simultaneous information on furan test results, moisture content; O₂, CO, and CO₂ levels; MeOH, EtOH, and the typical operating temperature (at least the top oil temperature). More complex multi-variable models could provide a more reliable prediction of transformer health, but they also increase the complexity of the analysis. Further investigations should be directed towards clarifying the molecular mechanisms underlying methanol generation during cellulose paper ageing. Specifically, future research should aim to identify the chemical pathways responsible for methanol formation with a focused exploration of the role of lignin decomposition. Furthermore, it should provide details on transformer design, construction, loading profile, and application to ensure accurate assessment and analysis. Also, similar relationships between cellulose paper ageing and the ageing indicators for conventional mineral oil have been reported in [39] to apply to Gas-to-Liquid (GTL) insulating liquids. However, further extensive research is needed in this regard, particularly for the emerging insulating liquids such as bio-based hydrocarbon and GTL. Authors in [96] present a comprehensive review of these two emerging insulating liquids.

Most of the current diagnostic techniques for monitoring insulation liquid conditions, which rely on the total acid number, are limited as techniques cannot differentiate between various acid types and their specific effects. Understanding the evolution of LMA is crucial for transformer fleet managers, as the condition of insulating paper directly determines a transformer service life. Therefore, developing new measurement techniques capable of distinguishing LMA would significantly improve the monitoring of the paper insulation system. Further research should explore the evaluation of the reaction rates of hydrolysis and transesterification processes. It is also important to investigate the impact of acids produced during the hydrolysis of esters. Furthermore, experimental and simulation observations have confirmed that variations in cellulose chain length do not result in significant differences in molecular conformation or physicochemical properties [52]. Therefore, cellulose chains of varying lengths can be effectively utilised for molecular and simulation modeling of cellulose paper.

Further study can focus on comparing the number of hydrogen bonds in insulation models under different ageing conditions. Given that hydrogen bonds play a crucial role in maintaining the structural integrity of cellulose fibres, further studies should investigate their behaviour under varying thermal and electrothermal stress. Specifically, researchers can analyse how increasing temperature affects the hydrogen bond network and contributes to the degradation of cellulose and insulation paper. Also, exploring the combined effects of thermal and electrical fields on hydrogen bond stability could provide deeper insights into insulation ageing mechanisms, ultimately guiding the development of more resilient insulation materials.

It is recommended to adopt the multi-characteristic comprehensive assessment method over the single-characteristic approach due to its advantages in comprehensiveness and accuracy. However, it is important to recognise the challenges associated with this method. The simultaneous analysis of multiple characteristic indices, such as MeOH and 2FAL, increases the complexity of data processing. Moreover, accurately determining the weights of different characteristics is crucial, as each characteristic has a varying impact on the DP value. Therefore, a careful experimental and theoretical analysis should be conducted to determine the appropriate weights for each characteristic to ensure precise and reliable assessment outcomes. Furthermore, it is observed that the DP profile in both axial and radial directions indicates that paper degradation follows thermal gradients, with lower DP values found in the top and inner windings due to higher operating temperatures. It is recommended to increase the number of transformers investigated to establish a more reliable correlation between the operating conditions of in-service transformers and the furan test results. This would enhance the effectiveness of furan testing as a tool for determining the degree of degradation in insulation paper.

According to findings in [97,98], insulating paper derived from bamboo, after undergoing a purification process, which involves hydrochloric acid and magnesium salt treatment, has demonstrated the enhanced physicochemical, mechanical, and insulation properties of bamboo paper. These enhancements include better impregnation performance, increased cellulose crystallite size, and improved tensile and tearing indices. Also, bamboo paper exhibits good thermal stability, making it suitable for use in high-temperature environments. Therefore, it is recommended to explore and optimise the use of bamboo paper as a low-carbon substitute for wood insulating paper. However, researchers should pay attention to the fact that bamboo paper still lags behind wood paper in key performance areas, such as tensile strength, breakdown strength, and DP. To address these limitations, future research should focus on optimising pulp modification, adjusting fibre orientation, and refining the purification process to enhance the properties of bamboo paper. These improvements could lead to a more competitive, sustainable alternative to traditional wood insulating paper, particularly in applications requiring high electrical strength and durability.

It is recommended that alternative diagnostic techniques be explored for assessing the ageing condition of Nomex insulation, as traditional cellulose-based indicators, such as the DP, furanic derivatives, and methanol and ethanol analysis, may not be as specific to it as they are to cellulose paper. However, tensile strength testing is the only reliable method, but it requires the physical extraction of insulation samples, which is costly and impractical. Currently, there is no specific and widely recognised chemical marker for directly assessing the degradation of Nomex paper in transformers. However, by combining various indirect markers like dissolved gases, oil acidity, furans, and monitoring techniques such as DGA, moisture content, and thermal analysis, the condition of the aramid paper may be assessed in conjunction with the overall condition of the transformer insulation system. For more precise assessments, advanced techniques like spectroscopic analysis or polymer-specific markers could be used. Therefore, research should focus on identifying potential chemical markers in the insulating liquid that could serve as indirect indicators of Nomex degradation, enabling non-invasive and cost-effective condition monitoring.

Furthermore, regardless of the type of paper, implementing sensor-based real-time monitoring as reported in [99,100,101] could provide continuous assessment and better capture chemical marker variations after insulating liquid replacement. Integrating these advanced sensing technologies could enhance measurement precision and enable early detection of insulation degradation.

9. Conclusions

The degradation of cellulose-based insulation in transformers is a complex process influenced by multiple factors, including thermal, electrical, and chemical stress. This study has reviewed the primary ageing mechanisms, which are pyrolysis, hydrolysis, and oxidation, highlighting their effects on the molecular structure of insulation paper. The role of dissolved gases and chemical markers, particularly 2FAL and methanol, has been discussed as diagnostic tools for assessing insulation ageing. Furthermore, the advantages of natural esters over mineral oil in preserving paper integrity have been explored, with evidence supporting their ability to reduce hydrolytic and oxidative degradation through enhanced moisture management and molecular interactions. While existing DP estimation models provide valuable insights into insulation ageing, challenges remain in accurately correlating chemical marker concentrations with real transformer conditions. Future research should focus on refining ageing models and developing more precise online diagnostic techniques to enhance transformer reliability. To address these challenges, utilities and transformer manufacturers are working on more sophisticated monitoring systems that combine multiple diagnostic techniques, and there is ongoing research into new materials and methods that could provide more direct and reliable insights into paper insulation conditions. By advancing our understanding of insulation degradation, the industry can implement better maintenance strategies to extend the operational life of transformers.