Correlation of tangible quality parameters of vegetable-based transformer fluids

Abstract

Due to the inherent environmental footprint of petroleum derived transformer fluids, the power industry is gradually exploring the potential of vegetable oils as alternatives. The impetus comes mostly from vegetable oils renewability and their inherent biodegradability.

However, the major drawback in the use of vegetable oils as dielectric fluids is their lower oxidative stability and higher kinematic viscosity compared to mineral oils. The results obtained clearly demonstrate the correlation between spectroscopic data induction time, kinematic viscosity, acid value, and peroxide value. Quantitatively, the absorption frequencies of functional groups in vegetable oil transformer fluids that can be correlated to the mentioned quality parameters show noticeable changes with aging/oxidative degradation. The study also demonstrates the utility of integrating spectroscopic data to understand trends in induction time and kinematic viscosity of oil samples heated under transformer service conditions.

Graphical abstract

1. Introduction

The transformer is the single most important component of a power distribution system and represents the largest portion of capital investment in electrical power transmission and distribution systems [1]. In its simplest form, the electrical transformer consists of primary and secondary turns [2] that enable a voltage step up or step down in accordance with the anticipated power distribution needs. To mitigate power losses due to eddy current in transformer windings, metal cores of transformer windings consist of plates separated by a paper insulator [3]. Statistical analyses show that the average age of a transformer on the grid is 35 years while under ideal conditions they are expected to operate for 30–40 years, with industrial transformers having a life expectancy of just 20–25 years [3]. For a power transformer to meet its operational service lifespan, certain service conditions must be met. For instance, the paper insulator must not undergo significant deterioration due to thermal effect, which requires that transformers are efficiently cooled.

To minimise thermo-degradation of paper insulators, liquid-filled transformers come in two principal ratings. There is the 55° Celsius rating, and the 95° Celsius rating, all with respect to the maximum internationally acceptable ambient temperature of 40 °C [4]. A study shows that for every 8° Celsius increase in the temperature of the transformer winding with regard to its rating leads to a 50% decrease in the anticipated lifespan [5]. Therefore, efficient cooling is necessary to achieve the service life. To achieve this objective, the power industry started using petroleum based mineral oils as dielectric/transformer fluids. In this regard, petroleum based oils have been used as liquid dielectrics for transformers as far back as [6]. In 1890, five years after the transformer invention, Brown Bovery & Cie (BBC) in Germany was the first company to apply minerals oil as transformer insulating and cooling media [7].

These low viscosity oils served the purpose of providing excellent insulation when impregnated into paper or other solid dielectrics. Besides, they also served as excellent heat transfer media for the removal of heat resulting from power loses. However, due to their widely acknowledged environmental footprint, the power industry has been looking for environmentally friendly alternatives. For instance petroleum based transformer fluids have low biodegradability [8]. In addition to this imminent environmental problem, petroleum based dielectric fluids are non-renewable in nature and could fail to meet demand in the event of global crude oil depletion.

Novel dielectric fluids for the power industry emphasize on liquid materials that are guided by multiple factors such as strict environmental regulations, fire safety and economic considerations. With these factors in mind, natural esters or vegetable-oil based transformer fluids are increasingly replacing mineral oil-based products in the market place [9] given that their dielectric constants rival those of mineral oils [10]. However, the major shortfalls of natural esters are their lower Oxidative Stability Index (OSI), which is evidenced by their peroxide values. They also have higher viscosity in addition to high acid values which lower Oil Quality Index (OQI) [11]. All these physical parameters depend on the molecular structure of natural ethers. Therefore, the viscosity, peroxide value and acid value can be correlated to the concentration of double bonds in the triglyceride molecule [12].

In view of its high potential as a substitute for transformer and lubricating fluids, several studies have been devoted to studying its oxidative potential, using different methods. For instance, Symoniuk et al. [13] have compared the oxidative stability of linseed oil, using the pressure differential scanning calorimetry (PDSC) and Rancimat methods under different thermal loading conditions. They reported that the PDSC method is more convenient for the determination of the induction time of linseed oils compared to the Rancimat method, and that the reaction rate coefficient increased with rising temperature during measurement by both methods. In another study [14], oils obtained from peanut were evaluated in terms of edible properties and suitability for biodiesel production. The study targeted free fatty acids contents, oleic acid contents and oxidative stability characteristics, with a low free acid and a high oleic acid content being common features. A poor oxidation stability characterized by high peroxide value and high spectrophotometric indices were also reported. Moreover, the effect of the extraction methods, based on stirring and Soxhlet, and four different solvents: acetone, chloroform, ethyl acetate and petroleum ether on the physiochemical properties of tomato seed oil has been studied by Giuffrè et al., [15]. The authors reported that the Free Fatty Acid Value, the p-anisidine value and the oxidative stability index were different for different extraction method. They also, found that the Soxhlet extraction always produced the tomato seed oil with the highest Totox index. Spectrophotometric characteristics described a non-oxidized vegetable oil. Moreover, the effect of the extraction methods, based on stirring and Soxhlet, and four different solvents: acetone, chloroform, ethyl acetate and petroleum ether on the physiochemical properties of tomato seed oil has been studied by Giuffrè et al. (2017) [15]. The authors reported that the Free Fatty Acid Value, the p-anisidine value and the oxidative stability index were different for different extraction method. They also, found that the Soxhlet extraction always produced the tomato seed oil with the highest Totox index. Spectrophotometric characteristics described a non-oxidized vegetable oil. Biological properties, such as iodine value, oxidative stability index, antiradical (2, 2-diphenyl-1-picrylhydrazyl radical, DPPH) activity, and phenol content also decreased when time and temperature increased [16].

Vegetable oils are natural esters of glycerol and fatty acids with two important chemical bonds, namely the ester bond and the ethylenic double bonds in unsaturated fatty acids [17]. This latter bond is susceptible to oxidative attach through the attack of reactive molecular oxygen species [18] leading to the formation of lipidroperoxides which can propagate free radical reactions [19]. In addition, tertiary oxidation products such as short chain free fatty acids can be present in vegetable oils [20]. Besides, the method of extraction of vegetable oils determines the concentration of free fatty acids [21]. For instance, in the extraction of vegetable oils where the slurry is heated to float the oil and its eventual recovery, hydrolytic reactions [22] can lead the formation of more free fatty acid compared to the cold press method and this reaction forms the basis for the commercial production of fatty acids from seed oils [23]. Therefore, the content of free fatty acids in vegetable oils is an indicator of the extent to which degradation has taken place and it will determine the quality of a biobased transformer oils, such as the kinematic viscosity and acid value as specified by the [24] and the oxidative stability [25]

Different portions of the electromagnetic spectrum have different effects on matter. The infrared spectrum incites vibrational deformation [26] in chemical bonds that enables it to be used for analysis of chemical structure. Attenuated Total Internal Fourier Transform Infrared spectroscopy is a versatile non-destructive chemical analytical technique that has found increased use in a broad spectrum of industries [27] and academia [28]. In this work, we correlate kinematic viscosity, peroxide value, acid value and oxidative stability data of vegetable-based transformer fluids to spectroscopy data measurements of AireSun Global transformer oil samples. We also determined experimentally these quality parameters using internationally certified approaches. The study demonstrates the utility of integrating spectroscopic data to understand trends in induction time and kinematic viscosity of oil samples heated under transformer service conditions.

2. Background theory: relationship of composition and molecular structure to the quality and physical properties of triglycerides

Vegetable oils are natural esters of glycerol and saturated and unsaturated long chain fatty acids, occurring as triglycerides, diglycerides and monoglycerides and free fatty acids (FFA) [29,30] The Acid Value (AV) of vegetable oils is a quality indicator that is directly correlated to the concentration of free fatty acid which has two origins. First, lipases are ubiquitous in nature and are produced by several plants that catalyze the hydrolysis of seed oil to FFAs [31] Second, much of vegetable oil is produced by the cold press method [32] followed by distillation where a solvent is distilled off. Therefore, hydrolysis leading to the cleavage of ester bond can produce FFs [33] as given in Eq. (1).

In addition to AV, the peroxide value is another quality indicator of lipids. Peroxides form as a result of lipid peroxidation reaction which involves the reaction of reactive oxygen species called singlet oxygen with the electron rich centers of double bonds of fatty acids (FA) in triglycerols. In this regard, photosensitized oxidation initiates oxidative deterioration of vegetable oils [34,35] where chlorophyll-like pigments present in oils act as sensitizers, absorbing visible light to produce hydroperoxides in unsaturated fatty acids [35]. This reaction has been categorized into two major classes, namely Type I and Type II [36]. In Type I reaction, free radicals are produced by interaction of the excited sensitizer with a substrate. In the Type II process, the excited sensitizer produces singlet oxygen by transferring excitation energy from the sensitizer to the thermodynamically stable triplet oxygen. The singlet active oxygen molecule reacts with olefinic double bonds to produce hydroperoxides [37]. Fig. 1 sums up the reaction mechanism.

Fig. 1.

Formation of peroxy radical.

This reaction constitutes oxidative damage to polyunsaturated vegetable oils [38]. The O–O bond of peroxides easily breaks, producing free radicals, which can initiate free radical reactions in vegetable oils to generate secondary oxidative products [39].

Triglycerides also contain secondary plant metabolite that control quality. Tocopherols, collectively known as vitamin E, are important secondary metabolites that are lipophilic antioxidants and exclusively synthesised by photosynthesis [40]. They fall in the class of phenolic antioxidants that function as hydrogen donors [41] to quench free radical, thereby boosting oxidative stability and improving lipid quality.

Carotenoids are highly pigmented compounds found in various foods such as vegetable oils that scavenge free radicals in photosensitive induced oxidation of lipids, which is another physicochemical mechanism that impacts the quality of vegetable oils [42]. Because of their highly conjugated polyene structures, these compounds act as singlet oxygen quenchers and radical scavengers and play an important role in the prevention of oxidative stress [43]. Therefore, the concentration of carotenoids in vegetable oils will determine their quality. Consequently, the peroxide value, FFA, tocopherols and carotenoids will also have significant bearing on the quality of vegetable oils in general as will the concentration of double bonds, given that they are the targets of Reactive Oxygen Species (ROS), and the application of sophisticated analytical chemical techniques and wet chemistry to the determination of the above-mentioned quality parameters can lead to meaningful correlation of tangible quality parameters and the following section will be devoted to this task.

3. Methodology

3.1. Materials

Five transformer oil samples coded as T001, T002, T003, T004 and T005 were supplied by AIRESUN GLOBAL Ltd of Glace Bay, Cape Breton, Canada. All the samples were supplied in opaque bottles with no potential for photo-induced oxidative degradation [44]. All samples in their containers were stored in a refrigerator at 0 °C to avoid thermal induced degradation [45].

All reagents were laboratory grade. Isopropyl Alcohol (99.99%) was obtained from Burdick and Jackson of USA. Triethanolamine (99.4%) was obtained from Baker Analyzed, USA. Potassium Nitrate (greater or equal to 99.0%) was obtained from VWR Chemical, USA. Hydrochloric acid (37%) was obtained from Sigma Aldrich, USA. The solution for each experiment was prepared using these laboratory grade reagents.

3.2. Experimentation

3.2.1. Acid value determination

The procedure used is greener, and it is based on the pH metric technique without titration [46] which is a modified version of the American Oil Chemists Society approach, where the presence of triethanolamine in the reagent causes rapid (within 1 min) and full extraction of free fatty acids from the test oil portion into the reagent to form an emulsion. In this procedure, two pH measurements, pH1 and pH2 are required for calculation. The reagent has the following composition: Composition of reagent: 0.2 M Triethanolamine + 50% vol. water + 50% vol. Isopropyl Alcohol. The standard reagent is first put in a beaker with a mass of potassium nitrate that will produce 0.02 M of this reagent. 8 mL standard acid solution of HCl (5*10⁻³ M) was added to the mixture and stirred to obtain homogeneity. An initial pH1 was then measured using the pH meter. Sample analysis was carried out by adding known masses of oil samples to the mixture followed by stirring. A final pH2 was then measured. The acid number is calculated based on the following Eq. (1) [46]:

$$AV=56.11\frac{N_{st}{V}_{st}}{\left({10}^{\Delta pH}\right)m}\left(\frac{mgKOH}{goil}\right)$$ (2)

In which $AV$ is the acid value [mgKOHg⁻¹], $N_{st}$ is the concentration of standard acid solution [M], $V_{st}$ is the volume of standard acid solution [cc], $m$ is the mass of oil [g] and $\Delta pH$ = pH1-pH2

pH1 is the pH of the reagent with the addition of the standard solution and pH2 is the pH of the mixture after the addition of oil sample.

3.2.2. Determination of peroxide values of oil samples

The procedure used for Peroxide Value (PV) determination is in accordance the IFRA method [47]. The process involves titration of 0.1 N solution of sodium thiosulfate solution until the deep blue color changes at the neutralization point. The reagent consists of 3 part of glacial acetic acid to 2 parts chloroform, concentrated potassium iodide solution, 1% starch solution as indicator, 0.1 N sodium thiosulfate solution as titrant and 3 g of oil sample. All reagents were laboratory grade acquired from Sigma Aldrich and used without any purification. Titration was conducted until the deep blue color turned purple. The International Scientific Committee of Ozone Therapy formula is given by Eq. (3) as [48].

$$PV=\frac{1000\left(V_1-V_0\right)c}{m}$$ (3)

where

• PV is peroxide value – mEq/kg

• V₁ is volume of sodium thiosulfate solution consumed in the titration of involving oil sample-cc

• V₀ is volume of sodium thiosulfate consumed in the blank titration that does not contain sample-cc

• C is the concentration of sodium thiosulfate solution (0.1 N)

• M is the mass of oil sample analyzed

In all cases, 3 g of oil sample was used. Because we did not have any idea about the peroxide values of oils, we decided to use 0.1 N solution of the titrant to be on the safer side [47].

3.2.3. Kinematic viscosity measurements

The kinematic viscosity of transformer oil samples were measured, using Cannon-Feske glass viscometer, based on the recommended standard of ASTM and IEEE, which recommends an internationally allowable maximum ambient temperature of 40 °C [49,50].

3.2.4. Spectroscopy method

The spectrometer used was the Nicolet Summit Attenuated-Total-Internal Reflectance-Fourier Transform-Infrared spectroscopy Transmission equipment manufactured by Thermoscientific. The detector is Lead Selenium (PbSe), the crystal type is diamond and the optical velocity is 0.4747. The accessories designed specifically for liquid samples were integrated to facilitate the acquisition of sample spectra in Absorbance mode. Usually, very high spectral resolutions are not needed to analyze solids and liquids samples and 4 cm⁻¹ is more than enough [51]. Therefore, following [52], we used a resolution of 4 cm⁻¹ and 16 scans were collected for both background and sample tests within the mid infrared region of the electromagnetic spectrum (400 cm⁻¹-4000 cm⁻¹).

3.2.5. Simulation of transformer oils under in-service conditions

Autoxidation of lipids requires them to be in radical forms. Generally, fatty acids or acylglycerols are in nonradical singlet states, and the reaction of fatty acids with radical state atmospheric ³O₂ is thermodynamically unfavorable owing to electronic spin conservation [53]. The hydrogen atom of the fatty acids in edible oil is removed, generating lipid alkyl radicals in the initiation step. Heat, metal catalysts, ultraviolet light and visible light can accelerate free radical formation of fatty acids or acylglycerols. To simulate the in-service performance of transformer oils, we heated the oil samples in a copper tube as a metal catalyst with a strip of Kraft paper as the cellulose insulators in transformer windings. The presence of cellulose also enabled us to study the effect of oil heating on insulator quality indicated by the presence of furfurals [54] from cellulose degradation.

3.2.6. Oxidative stability test

Oxidation stability is one of the most important quality indicators of edible vegetable oils. It measures their usefulness in technological processes as well as shelf life. The most reliable test is the Rancimat test. The Rancimat Accelerated Oxidation method consists of exposing an oil sample to a constant high temperature (between 50 and 200 °C), and airflow (between 1 and 25 L h⁻¹) to guarantee a sufficient oxygen supply to rapidly induce lipid oxidation. In this experiment, 3 g of the sample was heated in a reactor fitted with a tube that conveys gaseous reactants products to container containing deionized water with a conductivity meter. Air was pumped through the heated reactor that swept the gaseous products to the container of distilled water. Generally, low molecular weight carboxylic acids are produced and their dissolution in the deionized water produces ions which cause conductivity spike. Experimental data recorded were conductivity versus time. The VWR Air Flow Meter manufacture by Avantor was used in this study. Fig. 2 shows the experimental setup.

Fig. 2.

Schematic of the Rancimat test.

In all quantitative determinations involving peroxide value, acid number, kinematic viscosity and induction time, experiments were carried out in triplicates and the mean values for 3 determinations were reported.

4. Results and discussion

4.1. Infrared spectroscopy data analysis

The quality of a transformer fluid depends on its ability to circulate heat from hotspots to prevent accelerated deterioration of the cellulose insulator, and to also act as an efficient dielectric fluid by storing charges in alternating electromagnetic fields [55]. The efficient cooling is linked to the kinematic viscosity of the oil. The dielectric properties and dipole moment of vegetable oils have been found to depend on the Free Fatty acid content [56]. Therefore, knowledge of the concentration of double bonds or the degree of unsaturation, which determines the kinematic viscosity and the extent to which autooxidation can occur [57] is essential and can be obtained from vibrational spectroscopy data. The infrared spectra of a molecule are distinguished by two regions, the fingerprint region (400–1500 cm⁻¹) characterized by peaks that result from bending molecular vibrations and the functional group region that is characterized by stretching molecular vibrations (1500 cm⁻¹–4000 cm⁻¹) [58]. The fingerprint region characterized the type of molecule while the functional group region carries information about the type of identifiable bonds. In this regard, the fingerprint regions of samples, T001, T002, T003, T004, T005 and those of simulated oils reflects the library for OMIC paradigm regarding vegetable oils. The most distinct absorption band at 1743 cm⁻¹, regrading Fig. 3, Fig. 4, Fig. 5, Fig. 6, Fig. 7 can be assigned to the C=O stretching vibration of aliphatic esters [59]. The strong bands at around 2922 and 2852 cm⁻¹ can be ascribed to the asymmetrical and symmetrical C–H stretching vibrations of CH₂ methylene groups, while the band at around 1157 cm⁻¹ may be assigned to the stretching of the C–O bonds of aliphatic esters or CH₂-bending vibration [60]. The spectra for all 5 oil samples show vibration peaks at 3008 cm⁻¹ which corresponds to the (CH = CH) olefinic double bond [61].

Fig. 3.

Spectra of fresh sample T001.

Fig. 4.

Spectra of fresh sample T002.

Fig. 5.

Spectra of fresh sample T003.

Fig. 6.

Spectra of fresh sample T004.

Fig. 7.

Spectra of fresh sample T005.

It has been shown that a shift of the band at 3006 cm⁻¹ related to olefinic (CH=CH) double bonds reflect the degree of unsaturation. In this regard, a shift of this peak maximum to lower wavenumbers suggests a higher degree of saturation of olefinic double bonds [62]. A shift of the peak maximum for olefinic (CH=CH) double bond from 3008 cm⁻¹ to 3006 cm⁻¹ is observed for heated oils.

The double bond vibration at 2006 cm⁻¹ has absorbance 0.02, 0.017 and 0.011 for T001, T002 and T003 respectively, corresponding to absorbance 95%, 96% and 97% (Fig. 3 through Fig. 5) respectively. The cis-olefinic (>C=C<) stretching vibration, abbreviated as v(C=C) occurs at 1654 ⁻¹ cm [63] but this has not been identified. The double vibration corresponding to the 2006 cm⁻¹ for T004 (Fig. 6) has an absorbance of 0.003, corresponding to transmittance of 99%. The absorbance for T005 (Fig. 7) at this frequency is 0.007, corresponding to a transmittance of 98%. Transmittance was calculated using the formula in Ref. [64].

The absorption of the O–H group of carboxylic acids present in lipids does not usually appear as a sharp peak. Instead a broad band is observed because of the vibrational by hydrogen bonding network, with the degree of broadening being dependent on the degree of hydrogen bonding [65]. The frequency of the O–H vibration is visible between 3400 cm⁻¹ and 3600 cm⁻¹ for all samples except T002 and T003, but these two oils samples show distinct absorptions at 1094 cm⁻¹ and 1095 cm⁻¹ respectively, corresponding to bending vibrations. Consequently, intramolecular hydrogen bonding may have caused the absence of OH peaks at the 3400 cm⁻¹ due to frequency shift as can be seen from the spectra where the peaks become excessively broader towards higher wave numbers thereby appearing flat.

4.2. Correlation of tangible oil parameters to spectroscopic data

4.2.1. Acid value (AV)

Free fatty acids are naturally not found in vegetable tissues while growing but are generated enzymatically as soon as the seed is harvested. Thus, a lipidomic analysis revealed that during natural and artificial aging of Arabidopsis seeds, levels of several diacylglycerols and free fatty are generated [66]. The Free Fatty acid becomes extracted into the oil during extraction. Also, steam stripping of vegetable oils [67] during production can result in hydrolytic production of Free Fatty Acids. The extent of acidification of vegetable oils by these processes is measured as the acid value, which is the milligram of potassium hydroxide required to neutralize 1 g of the oil [68]. Table 1 shows the acid values of samples determined using the modified American Oil Chemists Association method [46]. The table shows that T001 has the lowest Acid Value followed by T003. The data in Table 1 is correlated to infrared spectroscopy data as follows:

Table 1.

Acid Values of samples.

Oil	Acid Number	Oil Quality Index
T001	0.034	274.41
T002	0.043	424.65
T00 3	0.041	285.12

Lipid peroxidation is a chain reaction initiated by hydrogen abstraction or via the addition of a reactive oxygen species/radical, resulting in the oxidative damage of polyunsaturated fatty acids (PUFA). Therefore, since polyunsaturated fatty acids are more sensitive than saturated ones, it is obvious that the activated methylene (RH) bridge associated with double bonds represents a critical target site. Consequently, on the basis of spectroscopic data in this study, there is more pronounced peroxidation of Free Fatty acids in T001 compared to the rest. Moreover, Yoshida [69] has demonstrated the influence of fatty acids of different unsaturation in the oxidation using microwave irradiation technique. The implication is that T001 will have a lower AV.

4.2.2. Peroxide value (PV)

Autoxidation of vegetable oils is a free radical reaction involving oxygen that leads to deterioration and rancidity with off-flavours and off-odours. Peroxide value, the concentration of peroxide in an oil or fat, is useful for assessing the extent to which auto-oxidation has occurred in the oil. Hydroperoxide occurs through the following oxidative attack on fats/oils given by Eq. (4) [70]:

R-H + O2→ROOH (4)

The determination of peroxide value relies on analytical chemical procedures that reflect the following reactions (Eq. (5) through Eq. (6)):

• Iodine generation step

Ki + CH₃COOH→HI + CH₃COOK–K⁺ (5)

Reaction of Hydroperoxide with hydrogen iodide for the generation of free iodine

ROOH + 2HI→H₂ + O + I₂ + starch as indicator (6)

Fig. 8a–d shows the snapshots of the different stages of the titration method used in PV determination.

Fig. 8.

a shows the reaction of potassium iodide with glacial acetic acid, b shows deep blue color after adding starch indicator, c shows purple color after neutralization point and d shows color of system during blank titration step. (For interpretation of the references to color in this figure legend, the reader is referred to the Web version of this article.)

Table 2 shows the peroxide values of samples reported as milli-equivalent per kilogram of samples with their corresponding Transmittances of double bonds. T001 has the highest PV, reflecting its higher concentration of unsaturation as determined by infrared spectroscopy data. T004 has the lowest PV of 18.21 mEq/kg followed by T003 with a value of 71.19 mEq/kg. T002 has the highest Transmittance corresponding to the lowest concentration of double bond, which reflects its lower PV compared to T001. Compared to T003, one would expect T002 to have a higher PV but the opposite is seen in Table 2. The reason for the opposite observation can be due to differences in the concentration of antioxidants and this aspect will be considered at the appropriate section. Also, based on Transmittance values, one would expect T004 to have a lower PV and this is seen in the table.

Table 2.

Peroxide and FFA values of oil samples.

Oil Sample	Mass of oil analyzed-g	Peroxide Value-mEq/kg	Transmittance-for double bond-%	Absorbance of Trans-fatty acids	Absorbance of FFA
T001	3	312.91 ± 1.9	95	0.05	0.017
T002	3	259.93 ± 0.44	96	0.027	0.008
T003	3	71.19 ± 0.54	97	0.015	0.007
T004	3	18.21 ± 0.33	99	0.005	0.018
T005	3	140.73 ± 0.43	98	0.003	0.013

Trans unsaturated fatty acids are less oxidizable than cis unsaturated fatty acids [71]. Trans-fatty acids show bending vibration at 966 cm⁻¹ [72] (See Appendix 1), which is due to the C–H out-of-plane deformation band, which is uniquely characteristic of isolated double bonds with trans configuration while conjugated double bonds absorb near 985 and 945 cm⁻¹ and near 990 cm⁻¹ [73].

In Fig. 3, Fig. 4, Fig. 5, Fig. 6, Fig. 7, the peaks for trans-fatty acid absorption occur as tiny ones that require tremendous effort to identify. Based on Table 2, T005 has more trans-fatty acids than T004 and will be expected to have a lower PV than T004, but the opposite is true and this can be due to differences in the concentration of antioxidants in both oils. Accordingly, the higher trans-fatty acid concentration of T001 reflects its lower AV in Table 1.

Recently, Betran et al. [74], confirmed that the only significant region for the quantification of Free Fatty Acid (FFA) is the adsorption band at 1711 cm⁻¹. Therefore, quantification of the FFA level has been achieved based on measurement of the peak height of the FFA carboxyl adsorption band at 1711 cm⁻¹, which was baseline-corrected by using 1850 and 1580 cm⁻¹ as anchor points. Table 2 shows the absorbance at 1711 cm⁻¹ for oil samples. Accordingly, T004 has the highest absorbance followed by T001 while T003 has the lowest value, but values of FFA acid absorption do not correlate with PV because higher concentration of these acids must reflect higher PV. Trans-fatty acids are generally resistant to oxidative degradation. The table shows that T001 has a higher concentration of Trans-fatty acid that could hinder lipid peroxidation reaction, but it has the highest PV. Moreover, T002 has the second highest trans-fatty acid absorption, but it has the second highest PV. Therefore, Trans-Fatty acid and FFA absorption data do not correlate to PVs of oil samples. However, transmittance values of oil samples in Table 2 pertaining to double bond absorption reflect the PV for T001 as well as for T002. Accordingly, T004 has the highest transmittance and will have the lowest concentration of double bonds and this reflects its lowest PV since double bond centers are more prone to oxidative attack.

Superoxo or peroxo complexes have been distinguished primarily on the basis of X-ray structural data (O–O) bond distance and vibrational spectroscopy (O–O) frequency [75]. In this regard, compounds with an (O–O) bond length of ≈1.4–1.5 Å and ν_{OO (O}–_O vibration frequency) between ≈800 and 930 cm⁻¹ are designated as peroxides (See Appendix 1). On the basis of this spectroscopic structural elucidation, Table 3 shows the infrared absorption of the (O–O) frequency obtained at 900 cm⁻¹ which in this study occurs between 876 cm⁻¹. Based on spectroscopic data from Table 3, T001 has the highest absorbance of 0.04, which justifies its highest PV. T002 ranks second to T001 with absorbance of 035 and a PV of 259.93 mEq/kg, the lowest absorbance for O–O vibration is 0.01 and this corresponds to its lower PV compared T001 and T002. The absorbance peaks for T004 and T005 are not visible.

Table 3.

Absorbance of O–O vibration and Peroxide Value.

Oil Sample	Absorbance for O–O Frequency	Peroxide value-mEq/kg
T001	0.040	312.91 ± 1.9
T002	0.035	259.93 ± 0.44
T003	0.01	71.19 ± 0.54
T004	–	18.21 ± 0.33
T005	–	140.73 ± 0.43

4.2.3. Kinematic viscosity

Table 4 shows kinematic viscosity of oil samples at the universally acceptable ambient temperatures of 40 °C and 100 °C, in accordance with ASTM 944 standard and the corresponding absorbances at the 3006 to 3007 cm⁻¹, which correspond to the degree of unsaturation. Based on the table, T001 has the highest absorbance corresponding to the highest concentration of double bonds. Therefore, T001 has the lowest kinematic viscosity at the two standard temperatures. Generally, greater the degree of unsaturation, lower is the melting point because in unsaturated oils, the presence of these double bonds renders the hydrocarbon chain less straight, which weakens the strength of the van der Waal’s forces because there is less contact between the chains. Therefore, the greater the degree of unsaturation, the lower is the viscosity.

Table 4.

Kinematic viscosity of oil samples at standard temperatures.

Oil Sample	Kinematic viscosity-cSt at 40 °C	Kinematic viscosity-cSt at 100 °C	Absorbance at 3006-3007 cm−1
T001	5.17 ± 0.05	1.91 ± 0.12	0.020
T002	41.4 ± 12	6.28 ± 0.08	0.017
T003	44.82 ± 0.11	1.72 ± 0.05	0.003
T004	14.40 ± 0.34	5.00 ± 0.06	0.01
T005	18.50 ± 0.25	5.82 ± 0.05	0.007

Rodrigueset al. [76], have observed that one double bond increases viscosity, whereas poly unsaturation caused a decrease in the viscosity. For instance, by comparing the viscosities of methyl stearate and methyl oleate, they confirmed that the presence of one double bond did increase the viscosity whereas polyunsaturation of two or three double bonds (methyl linoleate and methyl linolenate, respectively) reduced the viscosity. They argued that the presence of one carbon-carbon double bond in the structure of oleates resulted in stronger intermolecular interactions between the p electrons of the double bonds, which was possible because the spatial geometry of the cis configuration inherent in the one double bond of the oleate, which still allowed a close packing between the molecules. Based on this molecular level understanding, Table 6 can be used to correlate spectroscopic data to trend in kinematic viscosity as follows:

Table 6.

Interfacial Tension and Density values.

Oil sample	Interfacial Tension mN/m	Minimum value at 25 °C- mN/m	Absorbance of FFA	Density (g/mL)
T001	9.33 ± 0.11	40	0.017	0.879 ± 0.001
T002	18.26 ± 0.13	40	0.008	0.914 ± 0.002
T00 3	11.69 ± 0.37	40	0.007	0.909 ± 0.001

T001 has the highest absorbance of double bonds and should have the lowest values of kinematic viscosity at the standard temperatures, which indeed is the case. T002 has a higher absorbance than T003, which reflects values of kinematic viscosities of these samples. Also, T004 has a higher absorbance compared to T005 which reflects values of kinematic viscosity. T004 ranks third in absorbance for double bond concentration, which justifies its value of kinematic viscosity as seen in Table 4.

4.2.4. Oxidative stability tests

The Oxidation Induction Time (OIT) measurement is an accelerated thermal-aging test that provides a qualitative assessment of the material degradation. It is determined by the time interval to the onset of oxidation of a material at a specified temperature in an oxidized atmosphere. The onset of oxidation is detected by an abrupt increase in the sample’s evolved heat or temperature/conductivity. As thermal oxidation occurs, lower molecular weight carboxylic acids from oil are swept by air current into a container containing deionized water and ionization of the acid causes a spike in pH that is measured with time. Fig. 9 shows a typical plot of conductivity versus induction time for each oil sample while the resulting curves were evaluated with a graphical tangential procedure in the same way as found in the Rancimat method [77].

Fig. 9.

Example of plots for Induction time determination (Sample T001).

Based on the tangent method [78], Table 5 shows the induction times for oil samples. Accordingly, T001 has the shortest induction time followed by T003. On the basis of double bond concentration in Column 3 of the table, T001 should have the shortest induction time due to the higher susceptibility of double bond sites in lipids to oxidation and this anticipation is fulfilled. T002 has the next higher concentration of double bonds but it has a longer induction time compared to T003 with a higher concentration of double bonds. The observation relating to the comparison of T002 and T003 contradicts theory, but information on the concentration of Trans-Fatty acids as found in Column 4, which are resistant to oxidation reaction explains the trend. Once again, there is no meaningful correlation between trans-fatty acid absorbance and induction time, given that higher concentration of these acids should result in higher induction time. However, Column 5 of Table 5 shows that the higher concentration of carotenoid in T002, which as a free radical scavenger, has the potential to reduce oxidative degradation of T002 compared to T003. Based on carotenoid concentration, T003 must have a lower PV compared to T002.

Table 5.

Induction times of original oil samples.

Oil Sample	Induction time-min	Absorbance at 3006 cm−1	Absorbance of Trans-fatty acid	Absorbance for carotenoids at 960-965 cm−1
T00 1	22 + 0.36	0.020	0.05	0.048
T00 2	32 + 0.2	0.017	0.027	0.030
T00 3	26.3 + 0.10	0.003	0.015	0.007

4.2.5. Oil Quality Index

Transformer Oil Quality Index (OQI) is a quantitative assessment of its quality based on comparison of interfacial tension to acid value. It is defined as the ratio of interfacial tension between oil and distilled water to the acid value of the oil [79]. As transformer oil ages in service, oxidative products, such as carboxylic acids cause a decreased in oil-water interfacial tension while increasing the AV. Therefore, the ODI decrease with aging. Interfacial tension was measured using the Fisher Tensiomat Model 21 Manufactured by Fisher Scientific. Table 6 shows average values for 3 measurements for T001, T002, and T003. Values for T004 and T005 could not be measured due to insufficient sample volumes. From Table 6, T001 has the highest absorbance for FFA. Therefore, it has the lowest interfacial tension with water. Values of FFA for T002 and T003 do not reflect the difference in interfacial tension. However, density measurement has been shown to impact surface tension of liquids, where the higher the density the higher the surface tension [80]. The density of T001, T002 and T003 from Table 6 measured in accordance with ASTMD-941 are 0.879 gcm⁻¹, 0.914 gcm⁻¹ and 0.909 gcm⁻¹ respectively. On the basis of such experimental data, T002 is expected to have a higher surface tension compared to T003. From the molecular theory of interfacial tension, the interfacial tension between oil and water depends on their respective surface tensions and their polar and apolar contributions [81]. In this regard, the higher the value of the surface tension of oil, the higher will be the interfacial tension with water and this justifies the higher value of the interfacial tension between T002 and water compared to that between T003 and water.

Column 3 of Table 6 gives the minimum interfacial tension for good transformer oils based on ASTM standard. From the table, none of the oils meets the minimum requirement of 40 mN m which pertains to traditional transformer oils, such as minerals oils. Therefore, given the difference in the molecular structure of vegetable oils and that of mineral oils, the interfacial tension between mineral oil and water will be different from that of vegetable oils and water, and these standards cannot be applicable to vegetable oils as transformer fluids.

Table 7 for the OQI shows a lower value for T001, which is expected on the basis of spectroscopic and interfacial tension data. Based on density effect of interfacial tension, T002 has the highest OQI followed by T002 as found in the table.

Table 7.

Oil quality index.

Oil	interfacial tension	Acid Number	Oil Quality Index	Absorbance of FFA
T001	9.33 ± 0.11	0.034 ± 0.001	274.41 ± 0.15	0.017
T002	18.26 ± 0.13	0.043 ± 0.001	424.65 ± 0.16	0.008
T00 3	11.69 ± 0.37	0.041 ± 0.006	285.12 ± 0.54	0.007

4.3. Correlation of oil thermal loading characteristics to Spectral information

As transformer oils are heated in transformers, oxidative degradation leads to the formation of secondary oxidation products due to the cleavage of the double bonds to yield (di)carbonyl compounds (ketones and aldehydes) [82] and polymers [83]. In the presence of metals and water as catalysts, lipid hydroperoxides propagate radical reactions to produce these secondary oxidation products [84]. Such molecular alterations of vegetable oils lead to increased kinematic viscosity that can impair heat circulation by transformer fluids needed to prolong its life span. In this study, samples of T001, T002 and T003 were heated at 100 °C in a copper tube in the presence Kraft paper. The latter simulated the presence of the paper insulator in the transformer, which normally has a certain level of water [85]. By ASTM standards, bad oils look red to dark red in color. Therefore, oil samples were heater under maximal thermal loading conditions until red to dark red color was observed. This required heating for at least 12 h at 100 °C. Fig. 10(a–c) shows colors of heated oil samples. T001 had an intense dark color followed by T002 while the color for T003 was brick red. In all cases, the original fresh oil samples had light-yellow color, in accordance with ASTM color specification for good transformer oils.

Fig. 10.

Color of T001 after 12 h heating a, color of T002 after 12 h heating b and color of T003 after 12 h heating, c. (For interpretation of the references to color in this figure legend, the reader is referred to the Web version of this article.)

Fig. 11, Fig. 12, Fig. 13 show the spectra of simulated oil samples, while Table 8 shows measured kinematic viscosity and the corresponding spectroscopic data. Based on Table 8, there is a minimal change in viscosity of T001 after aging for 12 h compared to T002 and T003 where there are substantial differences. Therefore, under service conditions, T002 and T003 will undergo significant deterioration.

Fig. 11.

T00 1 Heated with Kraft paper.

Fig. 12.

T002 heated with Kraft paper.

Fig. 13.

T00 3 heated with Kraft paper.

Table 8.

Kinematic viscosity of simulated oils after 12 h of heating at 100 °C.

Oil	Kinematic viscosity-cSt	Absorbance for carotenoids
T001	3.11	0.048
T002	67.48	0.023
T00 3	16.38	0.037

The free radical scavenging properties of the oils may be due to the presence of carotenoids and phenolic compounds [86]. For instance, Song et al. [87] have evaluated the effect of different cooking methods on antioxidant content and radical scavenging activity of fresh and frozen sweet corn kernels by boiling, microwaving followed by determination of total carotenoid (TC) contents using spectrophotometric methods. Most intense signals observed in the spectra of isolated carotenoids due to wagging vibration of C–C can be seen at 965 and 960 cm⁻¹ for β-carotene and lycopene, respectively [88]. Table 8 shows the Absorbance of fresh oil samples corresponding to natural lipophilic carotenoid antioxidants prior to heating. Accordingly, T001 has the highest concentration of the natural antioxidant, which reflects its lowest viscosity. T003 has a higher concentration of carotenoids (0.037) compared to T002 (0.023). Therefore, T003 must have a higher antioxidant potential and this reflects its lower kinematic viscosity, and a higher antioxidant concentration in fresh oil samples is essential for lower kinematic viscosity which is the requirement for efficient cooling of transformers.

5. Conclusion

Vegetable oils are increasingly becoming substitutes for traditional petroleum derivatives as dielectric/coolants for transformers because of their promising biodegradability and potential for renewability. Accordingly, their quality parameters are fundamental issues in promoting their use. Vibrational spectroscopy is a fast and versatile chemical analytical technique that has a broad industrial application. In this study, we have correlated the relevant quality parameters of AireSum Global Ltd transformer fluids to infrared spectroscopic data. The following sum up the conclusion of the study:

• 1.Correlation of Free Fatty Acid absorbance data to quality parameters is more meaningful compared to correlation based on trans-fatty acid absorbance,
• 2.Absorbance data corresponding to trans-fatty acid and free fatty acids do not provide meaningful correlation to transformer oil properties.
• 3.T001 has the lowest induction time, but it has the highest resistance to oxidative degradation and this is confirmed by spectroscopic data related to carotenoid concentration which boots free radical scavenging resulting in lower kinematic viscosity,
• 4.Correlation of induction time and kinematic viscosity change after heating oil samples can best be done based on spectroscopic information relating to the initial concentration of antioxidant carotenoids free radical scavenger,

Author contribution statement

Adango Miadonye: Conceived and designed the experiments; Performed the experiments; Analyzed and interpreted the data; Contributed reagents, materials, analysis tools or data.

Mumuni Amadu; Conceived and designed the experiments; Performed the experiments; Contributed reagents, materials, analysis tools or data; Wrote the paper.

James Stephens: Analyzed and interpreted the data; Contributed reagents, materials, analysis tools or data.

Thomas O'Keefe: Analyzed and interpreted the data; Contributed reagents, materials, analysis tools or data.

Funding statement

This work was supported by Cape Breton University (80721).

Data availability statement

Data will be made available upon request.

Declaration of interest’s statement

The authors declare no conflict of interest.

Appendix 1. Absorbance of O–O frequency, ref. [66]