Dielectric Properties and Fire Safety of Mineral Oil and Low-Viscosity Natural Ester Mixtures in Various Concentrations

Abstract

This paper presents the results of testing the electrical and fire properties of mineral oil and low-viscosity natural ester mixtures. Properties such as breakdown voltage, relative permeability, dispersion coefficient, conductivity, flash and burn point, and lower heating values were investigated in different concentrations of mixtures of the two liquids, as well as for the base liquids. To ensure equal humidity levels, the prepared samples of mixtures and base liquids were conditioned under identical climatic conditions, resulting in samples with similar relative humidity (9 ± 3)%. The obtained measurement results for mixtures of the two fluids were related to the values obtained for the base liquids and analyzed in terms of changes in electrical properties and fire safety when used as insulating liquids in transformers. The presented results are useful for supplementing knowledge on the possibilities of using dielectric liquid mixtures in high-voltage power devices, with to the aim of using mixtures as alternatives to mineral oil.

1. Introduction

With the energy crisis and the intensification of environmental pollution problems, the application of technologies that help reduce pollution is attracting the attention of researchers [1]. This means that plant-based products continue to be a topic of interest for scientists around the world, including researchers working on the topic of power transformers. A power transformer is one of the key elements of the electricity transmission system. Its proper operation determines the proper functioning of power lines and transmission devices, as well as the transmission and distribution of electricity, determining the safety of an entire power system [2].

The relatively low cost of mineral oil means that most of the transformers operating in the power system are filled with this insulating liquid. It is estimated that several billion tons of transformer oil, which performs electrical insulation and cooling functions, is used worldwide [3]. However, international recommendations now seek to limit the use of these insulating liquids to protect the environment and prevent them from becoming unavailable. Furthermore, due to the poor environmental properties of mineral oil, such as low biodegradability [4], high toxicity [5,6], low flash point and fire point [7,8] as well as the fact that it is a non-renewable substance, scientists are still looking for alternative insulating liquids that could successfully replace mineral oil in the future. Electro-insulating liquids are still sought, which, unlike mineral oil, have a high tolerance to moisture, a higher degree of biodegradability, and a slow ageing rate [9]. The costs that the purchaser of a transformer must incur after purchasing, while operating, and at the time of recycling the used insulating liquid are also significant [9,10].

At the end of the last century, in the context of applications in the transformer insulation system, a new generation of ester-based environmentally friendly vegetable oils emerged [11]. These liquids have a much higher level of biodegradability [12] and fire safety [13] than traditional mineral oils. This fact makes transformers filled with vegetable oils safer for use in areas with a high population density and critical infrastructure, as well as in places where the risk of a possible failure could cause significant environmental contamination. The benefits of using vegetable oils appear to be just as great when one considers their electrical insulation properties and their compatibility with the other materials used in a transformer’s insulation system, such as paper or pressboard [14,15]. However, ester-based insulating vegetable oils are relatively expensive and have poor oxidation resistance, creating the need for hermetic transformer tanks, thus increasing the final price of the transformer [16].

Recently, in order to reduce the risk of transformer-related accidents and lower potential costs resulting from the use of appropriate fire protection systems and insurance, some managers have decided to modernise transformers previously designed for operation with mineral oil to adapt them for use with natural esters [17]. The process of replacing the insulating liquid currently used in the transformer (in this case, mineral oil) with a new liquid, such as a natural ester, is always associated with the contamination of the new dielectric liquid [18], which then leads to the formation of a mixture of both liquids (mineral oil and ester). The residual amount of mineral oil remaining in the transformer despite rinsing (oil in the pores of solid insulation, windings, core, on the bottom and walls of the tank) can differ depending on the transformer, amounting to 4–7% [19], 10% [20,21], and even 20% of the total volume [22]. Furthermore, some sources in the literature report that up to six months after retrofilling, mineral oil residues can move from solid insulation to the new insulating liquid, which is the ester in this case [23]. Such an action means that the mixing ratio of the two fluids (mineral oil and ester) can change over that time, affecting the dielectric, thermal and physicochemical properties, not only of the fluid that fills the transformer, but also of the cellulose insulation it impregnates [13,24,25].

Currently, in the context of searching for alternatives to dielectric liquids for mineral oil, many new liquids and additives (including nanoparticles) that improve the properties of previously used liquids are being studied [26,27,28]. Furthermore, mixtures of mineral oil with natural esters seem to be good alternatives in terms of improving their properties in relation to mineral oil [29]. According to [5], natural esters are miscible with mineral oil in all proportions. The mixtures themselves are also becoming widely acceptable since the price of esters is several times higher than the price of mineral oil [30]. A dielectric liquid can be obtained using mixtures of mineral oil with natural esters as the resultant, which is characterized by better fire properties (flash point, fire point), a higher degree of biodegradation and an increased water solubility limit compared to pure mineral oil [31,32]. As a consequence, the electrical strength of the insulating system and the service life of the solid insulation increase [33,34]. Furthermore, two studies [35,36] showed that mixtures of mineral oil and natural esters containing 10% to 25% ester are characterised by about 60–70% better electrical strength than pure mineral oil. There are also studies in which the authors compare the distribution of the electric field in power transformers filled with natural esters and mineral oil [17]. The authors state that a change in permittivity of dielectric liquids affects the distribution of stresses withstood by solid and liquid insulation. It should be considered that, in the case of newly modernized transformers, i.e., those in which the dielectric liquid has been replaced relatively recently, the permeability of solid and liquid insulation is uncertain, as it depends on the constantly changing concentration of mixtures of both dielectric liquids (as a result of mineral oil residues accumulated in the pores of the solid insulation entering into the ester).

Ageing processes lead to changes in the values of dissipation factor and conductivity, making it possible to determine the criteria values, the exceeding of which poses danger to further operation of equipment. Furthermore, the knowledge of the permittivity and DF values or the conductivity of the liquid is very helpful during the analysis of the dielectric response of paper–oil insulation, which aims to determine the moisture content of the solid insulation with the use of DFR (dielectric frequency response) and PDC (Polarisation–depolarisation current) methods [37,38].

Another important feature of mineral oil and natural ester mixtures is their heat transport capacity, which is mainly determined by liquid viscosity. It is well known that mineral oil, regardless of temperature, is characterised by significantly lower viscosity values than natural esters [24]. However, if other liquid properties that determine heat transfer, such as thermal conductivity, are taken into account, then natural esters have more desirable values. Therefore, in the context of the transformer cooling system, the use of mineral oil and natural ester mixtures seems to be a good solution, as this causes the viscosity of the resultant, which will be lower than that of natural ester, and thermal conductivity, which will be higher than that of pure mineral oil [31]. There are also natural esters that have almost twice as much low viscosity and thermal conductivity, which is about 6% higher than a typical natural ester [39].

In terms of fire safety, the temperature of hot spots in transformers filled with mineral oil and natural esters should also be taken into account. Annex C of the IEC 60076-14 standard [40] provides guidelines on temperature rise limits for transformers whose insulation system consists of Kraft paper and esters. An interesting fact is that these limits are 90 K for the hot spot temperature, which, compared to the IEC 60076-2 [41] standard on temperature rises for an insulation system consisting of mineral oil impregnated paper (limit 78 K), is a value of 12 K larger. The standard of [40] is also used in the retrofilling process, in which mineral oil is replaced withnatural esters with high flash point, fire point and autoignition temperatures [5]—Class K liquids [42] according to [43,44].

In connection with the above, it seems appropriate to study and understand the behaviour and changes in the properties of insulating liquid mixtures, which can be an alternative to the currently used mineral oil. this paper considers the electrical and fire properties of mixtures of mineral oil and low-viscosity natural esters, which ultimately may translate into an even better efficiency of the used mixture. The available literature focuses only on mixtures of mineral oil and commonly used esters (synthetic and natural), whose viscosity is several times higher than that of mineral oil. Mixtures of various insulating liquids, their properties (physicochemical, electrical), and mixing ratios are still an object of interest for various researchers looking to find the optimal mixing ratio. The test results presented in this publication may be useful to transformer owners and researchers interested in improving the properties of transformers that were previously filled with mineral oil.

2. Materials and Methods

Mixtures of mineral oil (MO) and low-viscosity natural esters (LVNEs) were the subject of the tests. The mineral oil Nynas Draco (Nynas, Stockholm, Sweden) [45], commonly used in transformers, and a low-viscosity natural ester with NOMEX 970FLD (DuPont, Wilmington, DE, USA) [46] were used in this study. Both pure (unmixed) base insulating liquids and their mixtures, characterised by the following percentage share of MO in relation to the LVNE—100/0, 95/5, 90/10, 80/20, 60/40, 50/50, 40/60, 20/80, 10/90, 5/95 and 0/100—were prepared for the tests.

Because the base liquids used (MO and LVNEs) were characterised by different water solubilities, despite the fact that their level of relative humidity was the same, the relative humidity of the obtained mixtures was diversified. Therefore, the prepared samples were subjected to a conditioning process in a climatic chamber (MKF, Binder, Bretnig-Hauswalde, Germany) in order to achieve the same level of relative humidity of the tested mixtures. The conditioning of the base liquids and samples of mixtures was carried out until all samples reached the same level of relative humidity, i.e., (9 ± 3)%. Relative humidity was measured using Vaisala MMT338 capacitive probes (Vaisala, Vantaa, Finland).

The tests included measurements of selected electrical and fire properties of mixtures and base liquids, i.e., mineral oil (MO) and low-viscosity natural esters (LVNEs).

In terms of electrical properties, the breakdown voltage (according to the standard of [47]) and the dissipation factor, relative permittivity and conductivity (according to the standard of [48]) were tested. On the basis of the measurements of the dissipation factor and conductivity, the activation energy for the AC and DC processes was determined. The breakdown voltage was measured using breakdown voltage tester, type WPOT, 0,25/75 (No. 884176, TUR, Dresden, Germany). Tests of permittivity ε_r, dissipation factor DF and conductivity σ were carried out using:

• Test cell for liquid insulation (type: 2903 AB, no. 118482, TETTEX, Switzerland, Zurich).
• Temperature regulator (type 2965 AK, no. 119627, TETTEX, Zurich, Switzerland, range: 30–150 °C).
• Insulation diagnostic analyzer (type IDAX 300 AG—19072, no. 080106, Megger Sweden AB, Täby, Sweden).

In order to determine the fire resistance of dielectric mixtures and their fire safety, the flash point, fire point and net calorific values of the tested mixtures and base liquids were measured. The flash point was measured using a closed-cup system, in accordance with the standard of [49]. Furthermore, according to the standard of [50], flash point and fire point measurements were carried out in the open cup system. Due to the fact that the net calorific value of the mixtures linearly depends on their molecular composition, according to the standard of [51], only the net calorific values of the base liquids (MO and LVNEs) were measured. The net calorific values of the mixtures were analytically determined.

3. Results and Discussion

3.1. Electrical Properties

Figure 1 presents the breakdown voltage of LVNE/MO mixtures. The average values of the breakdown voltage obtained for each mixture are shown in black. The results of individual measurements are highlighted in white. The scattering of the results obtained does not differ from the scatter observed by other researchers [36]. The standard deviation for the mixtures for each concentration did not exceed 12.

Figure 1 shows that for low contents of one liquid in the other, (from 0 to 20%), the breakdown voltage BV decreases by several percent compared to the breakdown voltage of the base liquids (BV of MO = 69 kV, BV of LVNE = 60 kV). The breakdown voltage reduction can be observed regardless of whether an ester is added to the mineral oil or mineral oil to the ester. On the other hand, for concentrations close to 50%, an inevitable increase in the breakdown voltage can be observed. However, from the uncertainty of the measurement (green error bars in Figure 1) calculated using the standard deviation for the confidence level of 0.95, it can be concluded that the observed changes in the breakdown voltage depending on the composition of the mixtures are not statistically significant.

The tests of permittivity ε_r, dissipation factor DF, and conductivity σ of the mixtures were carried out for two temperature values: 25 and 90 °C. Such values were selected as the limits of the temperature range in which power devices with oil-and-paper insulation most often operate. In all of the tests, permittivity and dissipation factor measurements were made at 50 Hz sinusoidal alternating voltage, while conductivity was measured at DC voltage. Figure 2 shows the relative permittivity relationship of the ester content in the mixture (suitably with mineral oil and natural esters) at 25 °C and 90 °C. This relationship is linear. Estimated equations approximating the measurement results are characterized by high squared Pearson correlation coefficients:

$$R^2={\left(\frac{\sum _{i=1}^n(x_{m_i}-\stackrel{-}{x_m})(x_{c_i}-\stackrel{-}{x_c})}{\sqrt{\sum _{i=1}^n({x_{m_i}-\stackrel{-}{x_m})}^2}·\sqrt{\sum _{i=1}^n({x_{c_i}-\stackrel{-}{x_c})}^2}}\right)}^2,$$

(1)

where:

x_m—permittivity measured for a specific ester content in the mixtures;

x_c—permittivity calculated on the basis of the approximating curve for a specific ester content in the mixtures;

n—number of points for which measurements and calculations were carried out.

Figure 2. Relative permittivity of the low-viscosity natural ester/mineral oil mixtures depending on ester content.

Figure 2. Relative permittivity of the low-viscosity natural ester/mineral oil mixtures depending on ester content.

Figure 3 shows the dependence of the dissipation factor on ester content in the mixtures at 25 °C and 90 °C, exponential trend lines, and correlation coefficients. On the basis of the measured values of dissipation factors, the activation energy (A_e) was calculated using the Arrhenius equation:

$$A_e=\frac{k·ln\frac{{DF}_{T_1}}{{DF}_{T_2}}}{\left(\frac{1}{{273+T}_1}-\frac{1}{{273+T}_2}\right)}$$

(2)

where:

k—the Boltzmann constant;

DF_T1, DF_T2—dissipation factor values measured at 25 °C and 90 °C, respectively;

T₁, T₂—temperatures of 25 °C and 90 °C, respectively.

Figure 3. Dissipation factor of the low-viscosity natural ester/mineral oil mixtures depending on ester content.

Figure 3. Dissipation factor of the low-viscosity natural ester/mineral oil mixtures depending on ester content.

The activation energy determined on the basis of Equation (2) is shown in Figure 4. The changes are linear and described by the equation shown in the graph.

The vertical error bars represent the uncertainty of determining the activation energy. The uncertainty of determining the activation energy (ΔA_e) was calculated using the total differential method, which defines the maximum possible value of uncertainty. The uncertainty of determining the activation energy was calculated according to this formula:

$$\begin{gathered} ∆A_e=\left|\frac{\partial A_e}{\partial {DF}_1}\right|∆{DF}_1+\left|\frac{\partial A_e}{\partial {DF}_2}\right|∆{DF}_2+\left|\frac{\partial A_e}{\partial T_1}\right|∆T_1+\left|\frac{\partial A_e}{\partial T_2}\right|∆T_2= \\ =\left|\frac{k}{{DF}_1·\left(\frac{1}{{273+T}_1}\right.-\left.\frac{1}{{273+T}_2}\right)}\right|∆{DF}_1+\left|-\frac{k}{{DF}_2·\left(\frac{1}{{273+T}_1}\right.-\left.\frac{1}{{273+T}_2}\right)}\right|∆{DF}_2+ \\ +\left|\frac{k·ln\frac{{DF}_1}{{DF}_2}}{{{(273+T}_1)}^2·{\left(\frac{1}{273+T_1}\right.-\left.\frac{1}{{273+T}_2}\right)}^2}\right|∆T_1+\left|-\frac{k·ln\frac{{DF}_1}{{DF}_2}}{{{(273+T}_2)}^2·{\left(\frac{1}{{273+T}_1}\right.-\left.\frac{1}{273+T_2}\right)}^2}\right|∆T_2 \end{gathered}$$

(3)

where:

k—the Boltzmann constant;

DF₁, DF₂—dissipation factor values measured at 25 °C and 90 °C, respectively;

T₁, T₂—temperatures of 25 °C and 90 °C, respectively;

ΔDF₁, ΔDF₂—absolute error in the dissipation factor measurement;

ΔT₁, ΔT₂—absolute error in the temperature measurement.

The activation energy can also be determined at direct voltage by measuring electrical conductivity at different temperatures. Figure 5 shows the relationships between the conductivity of the mixtures depending on the ester content in the mixtures at 25 °C and 90 °C. This relationship, as well as the relationship between the dissipation factor and ester content in the mixtures, is exponential. Equations describing conductivity changes depending on the composition of the mixture are given in the graph.

The squares of the correlation coefficients presented in Figure 3 and Figure 5 were calculated on the basis of Formula (1), where the respective variables $x$ are the following:

x_m—dissipation factor and conductivity measured for a specific ester in the mixtures;

x_c—dissipation factor and conductivity calculated on the basis of the approximating curve for a specific ester in the mixtures.

Using a similar formula as that for the dissipation factor, the activation energy for the individual mixtures was determined on the basis of conductivity:

$$A_e=\frac{k·ln\frac{{\sigma }_{T_1}}{{\sigma }_{T_2}}}{\left(\frac{1}{273+T_1}-\frac{1}{{273+T}_2}\right)}$$

(4)

where:

σ_T1, σ_T2—conductivity measured at T₁ = 25 °C and T₂ = 90 °C.

The results of the activation energy calculation are shown in Figure 6. As with the activation energy determined by the dissipation factor, the relationship is linear. Differences in activation energy values determined using the dissipation factor and conductivity were previously observed and are caused by the different polarization processes that occur at a constant voltage and alternating voltage of 50 Hz [17,40]. In this case, error bars also represent the maximum uncertainty in determining the activation energy, and were determined on the basis of a similar formula:

$$\begin{gathered} ∆A_e=\left|\frac{\partial A_e}{\partial {\sigma }_1}\right|∆{\sigma }_1+\left|\frac{\partial A_e}{\partial {\sigma }_2}\right|∆{\sigma }_2+\left|\frac{\partial A_e}{\partial T_1}\right|∆T_1+\left|\frac{\partial A_e}{\partial T_2}\right|∆T_2= \\ =\left|\frac{k}{{\sigma }_1·\left(\frac{1}{{273+T}_1}\right.-\left.\frac{1}{{273+T}_2}\right)}\right|∆{\sigma }_1+\left|-\frac{k}{{\sigma }_2·\left(\frac{1}{{273+T}_1}\right.-\left.\frac{1}{{273+T}_2}\right)}\right|∆{\sigma }_2+ \\ +\left|\frac{k·ln\frac{{\sigma }_1}{{\sigma }_2}}{{{(273+T}_1)}^2·{\left(\frac{1}{{273+T}_1}\right.-\left.\frac{1}{{273+T}_2}\right)}^2}\right|∆T_1+\left|-\frac{k·ln\frac{{\sigma }_1}{{\sigma }_2}}{{{(273+T}_2)}^2·{\left(\frac{1}{{273+T}_1}\right.-\left.\frac{1}{{273+T}_2}\right)}^2}\right|∆T_2 \end{gathered}$$

(5)

where:

k—the Boltzmann constant;

σ₁, σ₂—conductivity measured at 25 °C and 90 °C, respectively;

T₁, T₂—temperatures of 25 °C and 90 °C, respectively;

Δσ₁, Δσ₂—absolute error in the conductivity measurement;

ΔT₁, ΔT₂—absolute error in the temperature measurement.

Figure 6. Activation energy determined on the basis of conductivity depending on the ester content in the mixtures.

Figure 6. Activation energy determined on the basis of conductivity depending on the ester content in the mixtures.

Knowing the dependence of activation energy on the concentration (Figure 4 and Figure 6), the dissipation factor and conductivity of the mixtures for any temperature can be determined using these formulas:

$${DF}_1={DF}_2\mathrm{e}\mathrm{x}\mathrm{p}\left(\frac{A_{ex}}{k}\left(\frac{1}{(273+T_2}-\frac{1}{(273+T_1}\right)\right),$$

(6)

$${\sigma }_1={\sigma }_2\mathrm{e}\mathrm{x}\mathrm{p}\left(\frac{A_{ex}}{k}\left(\frac{1}{(273+T_2}-\frac{1}{(273+T_1}\right)\right),$$

(7)

where:

${DF}_1$, ${\sigma }_1$—the dissipation factor and conductivity in temperature T₁;

${DF}_2$, ${\sigma }_2$—the dissipation factor and conductivity in temperature T₂;

A_ex—the activation energy of the mixture with LVNE x, which, in the case of calculations of the dissipation factor, can be seen in Figure 4 and, in the case of calculations of conductivity, in Figure 6.

3.2. Fire Properties

Table 1. Flash point, fire point and net calorific value results of base liquids and mixtures of mineral oil and low-viscosity natural ester.

Mixing Ratio		Flash Point (Closed Cup)	Flash Point (Open Cup)	Fire Point (Open Cup)	Net Calorific Value	Fire Classification
Mineral Oil	Low-Viscosity Natural Ester
(%)	(%)	(°C)	(°C)	(°C)	(MJ·kg−1)
100	0	154.5 ± 1	154.7 ± 1	160.7 ± 1	42.2 ± 0.6	O1
95	5	156.1 ± 2	156.1 ± 2	162.8 ± 1	42.0	O1
90	10	156.8 ± 1	157.4 ± 1	164.8 ± 1	41.8	O2
80	20	157.4 ± 1	159.4 ± 1	167.4 ± 1	41.5	O2
60	40	159.4 ± 1	163.4 ± 1	173.4 ± 1	40.7	O2
50	50	161.2 ± 1	165.4 ± 1	176.1 ± 2	40.4	O2
40	60	163.2 ± 1	168.8 ± 1	185.5 ± 1	40.0	O2
20	80	172.5 ± 1	190.8 ± 1	208.8 ± 1	39.2	O2
10	90	185.2 ± 1	211.5 ± 1	232.2 ± 2	38.9	O2
5	95	194.5 ± 1	222.2 ± 2	248.2 ± 2	38.7	O2
0	100	211.1 ± 1	238.2 ± 2	264.2 ± 2	38.5 ± 0.2	O2

and Figure 7 show the flash point and fire point results for the mixtures of mineral oil (MO) and low-viscosity natural esters (LVNEs), depending on the mixing ratio of the base liquid and both liquids. For comparison, the results of the base liquids tests are also presented. The flash point was determined using two methods: the closed cup method and the open cup method. When the obtained results are analysed, it can be seen that, with an increase in the LVNE content in the mixture, the flash point and fire point also increase. A significant increase in flash point and fire point is observed for mixtures in which the LVNE content is at least 80%. The flash point of the MO and LVNE mixture determined using the closed cup method increases by 11.7% (18.0 °C) compared to the flash point of pure MO. In turn, the flash point and the fire point, determined using the open cup method, increase by 23.3% (36.1 °C) and 29.9% (48.1 °C), respectively, relative to pure MO. Similar behaviour was observed when determining the flash point and fire point of mixtures of mineral oil and the commonly used natural ester Envirotemp FR3 [13], as well as a mixture of mineral oil and the synthetic ester, Midel 7131 [52]. In addition, mixtures that contain more mineral oil in their volume, according to the research results presented in [53], generate more combustible gases such as carbon monoxide CO, methane CH₄, ethene C₂H₄, ethane C₂H₆, and propene C₃H₆, which contribute to reducing the flash point and fire point. Moreover, according to the authors, the differences in the molecular structure and chemical behaviour of mineral oil and natural esters may affect the generation of gases in the analysed dielectric liquids.

With an increase in LVNE content in the mixture, the differences between the flash points determined by both methods become increasingly visible. This effect is caused by the accumulation of combustible gases emitted by dielectric liquids when using the closed cup method to determine the temperature. When determining the flash point using the open cup method, the generated gases are continuously released into the environment (they do not accumulate in the cup), which means that their amount is insufficient for ignition [13]. As a consequence, the flash point determined using the open cup method is higher in each of the considered cases.

There are also visible differences between the flash point and the fire point, which were determined using the open cup method. As the LVNE content in the mixture increases, these differences become increasingly visible. Similarly, as stated in the authors’ previous publication [13], these differences are due to the structure of the tested liquids. Mineral oils are heterogeneous liquids: they are a mixture of cyclic-branched and straight-chained heavy alkanes [5]. The structure of mineral oils causes light fractions to be released faster during heating, as they evaporate with greater intensity. Moreover, volatile substances in mineral oil are released at lower temperatures than in natural esters [54]. Hence, the differences between the flash point and fire point determined by the open cup method become visible.

Table 1 and Figure 8 show the results of the measurements of the net calorific value of base liquids (MO and LVNEs) and the calculated values of the net calorific value for MO/LVNE mixtures. According to the data presented in [7,13], since the net calorific value of the insulating liquid mixtures linearly depends on the proportion of their mixing, this value was measured only in the case of base liquids; for MO/LVNE mixtures, it was analytically calculated. According to the data presented in Table 1 and Figure 8, it can be seen that with the increase in the content of a low-viscosity natural ester in the MO/LVNE mixture, the net calorific value decreases. Differences in the net calorific value of the tested liquids and dielectric mixtures are related to the chemical structure of both base liquids [55,56]. The analysed dielectric liquids are characterised not only by different types of chemical bonds, but also a different number of these bonds. Each type of chemical bond has a different energy. As a consequence, differences in the calorific value of the analysed dielectric liquids become visible.

The behaviour of the insulating liquid in fires is a complex issue that requires taking into account many of its properties, e.g., flash point, fire point, net calorific value, heat release rate, oxygen index, toxicity of decomposition products, etc. Some of these characteristics are well-defined via the use of existing ISO methods. However, some of them are still not standardised at an international level regarding teir measurement methodology. The international standard IEC 61100 [57] defines a system for classifying insulating liquids based on characteristics that can be quantitatively measured using standardised methods. Therefore, to classify electrically insulating liquids in terms of fire in accordance with the aforementioned standard, the combustion temperatures and lower calorific value are used. This standard distinguishes three classes of division for insulating liquids according to the fire point: class O (if the fire point is less than or equal to 300 °C), class K (if the fire point exceeds 300 °C) and class L (if the insulating liquid does not have a measurable fire point). The standard also distinguishes three classes according to net calorific value: class 1 (if the net calorific value is greater than or equal to 42 MJ·kg⁻¹), class 2 (if the net calorific value is lower than 42 MJ·kg⁻¹ and greater than or equal to 32 MJ·kg⁻¹) and class 3 (if the net calorific value is less than 32 MJ·kg⁻¹).

The last column of Table 1 contains the fire classification of the liquids and dielectric mixtures analysed. On the basis of measurement results for the combustion temperature, it can be seen that all the analysed dielectric liquids and their mixtures qualify for class O. In turn, based on the net calorific value, only mineral oil and a mixture consisting of 95% MO and 5% LVNE value can be classified as class 1. Other mixtures, that is, mixtures consisting of at least 10% LVNE and at most 90% MO and pure LVNE can be classified as class 2. As a consequence, in terms of fire classification, class O1 includes pure mineral oil and a mixture consisting of 95% MO and 5% LVNE, and class O2 includes the other tested mixtures (consisting of at least 10% LVNE and 90% MO) and pure low-viscosity natural ester LVNE.

4. Conclusions

Mixing a poorly biodegradable mineral oil (about 10% biodegradable) with a highly biodegradable natural ester (nearly 100%) can provide an alternative to mineral oils that produce harmful environmental effects. This publication presents selected electrical and fire properties of mixed samples of mineral oil and low-viscosity natural ester. On the basis of the presented test results, the following conclusions can be drawn:

–

Mixing mineral oil with a low-viscosity natural ester can contribute to increasing their flash points and fire points. In addition, taking into account the determined values of the net calorific value, it is shown that the transition from O1 to O2 classification occurs at 10% content of low-viscosity natural ester in the mixture. Small deviations from this proportion can be expected due to small discrepancies in the compositions of the analysed basic insulating liquids that may occur in individual batches of dielectric liquids.

–

When the composition of the mixture changes, the permittivity changes linearly, from the value characteristic for one liquid in the mixture to the value characteristic for the other liquid; at the same time, slightly greater changes in permittivity are observed at lower temperatures.

–

Changes in the dissipation factor and the conductivity of mixtures with an increasing share of LVNE against the MO are exponential, while the activation energy changes linearly.

Knowledge of the permittivity, dissipation factor, conductivity, and activation energy of the mixtures is necessary in the case of using the dielectric spectroscopy method to assess the moisture content of cellulose insulation impregnated with such mixtures.

Further research is necessary to verify whether mineral oil and low-viscosity natural ester mixtures meet all the requirements for dielectric liquids used in high-voltage power devices. These tests could aim to measure other properties of mineral oil and natural ester mixtures that could determine the behaviour of insulating liquids during fires, e.g., heat release rate, oxygen index, and toxicity of decomposition products.

From the point of view of operation, it is very important to determine the nature of the ageing phenomena occurring in such mixtures. In particular, the effects of water, and low- and high-energy discharges on the ageing of liquids should be investigated. The research described above, although time-consuming, is very important. The problems of material ageing are always complex and require a multicriteria critical analysis, which will be of interest to the authors in the future.