Gassing Tendency of Fresh and Aged Mineral Oil and Ester Fluids under Electrical and Thermal Fault Conditions

Operational factors are known to affect the health of an in-service power transformer and to reduce the capabilities and readiness for energy transmission and distribution. Hence, it is important to understand the degradation rate and corresponding behavioral aspects of different insulating fluids under various fault conditions. In this article, the behavior of mineral oil and two environmentally friendly fluids (a synthetic and a natural ester) are reported under arcing, partial discharges, and thermal fault conditions. Arcing, partial discharges and thermal faults are simulated by 100 repeated breakdowns, top oil electrical discharge of 9 kV for five hours, and local hotspots respectively by using different laboratory-based setups. Some physicochemical properties along with the gassing tendency of fresh and aged insulating liquids are investigated after the different fault conditions. UV spectroscopy and turbidity measurements are used to report the degradation behavior and dissolved gas analysis is used to understand the gassing tendency. The changes in the degradation rate of oil under the influence of various faults and the corresponding dissolved gasses generated are analyzed. The fault gas generations are diagnosed by Duval’s triangle and pentagon methods for mineral and non-mineral oils. It is inferred that; the gassing tendency of the dielectric fluids evolve with respect to the degradation rate and is dependent on the intensity and type of fault.


Introduction
Alternative dielectric fluids for use in oil-filled apparatuses is a topic of concern for industry and utilities for decades. Ester dielectric fluids (both natural and synthetic esters) are being investigated for use in oil-filled transformers. The dielectric properties and rate of deterioration of ester fluids is comparable to those of mineral insulation oils used in power transformers [1]. The compatibility of these alternative fluids with solid insulants was investigated and reported possible longevity of cellulose insulants when used with ester dielectric fluids [2]. The pre-breakdown phenomena, workability, influence of temperature, compatibility with other transformer materials, and service experience were reported by various researchers [1,[3][4][5]. Several researchers also attempted to understand the regeneration aspects and end of life criteria of these insulating fluids [6]. From the extensive laboratory studies, IEEE Standards and CIGRE reports came to the acceptance of using these fluids in oil-filled transformers [7][8][9]. However, condition monitoring and diagnostic tools for ester-filled transformers is still a topic of high interest for the industry. Meanwhile, the possibility of ester fluids to be miscible with mineral insulating oils was a topic of research to understand the feasibility of retro filling existing units. Fofana et al. [10,11] reported the properties of various mixtures of mineral oil and ester fluids extensively. Later, U. Mohan Rao et al. [12,13] reported the long-term degradation behavior, degradation solid insulation in mixture insulating fluids. Nevertheless, ester fluids are being used in oil-filled transformers and various utilities across the globe started filling/retrofilling some new and existing transformers with ester liquids. There is a large scope for research including, but not limited to, the understanding of the behavior of these insulating fluids under different operating conditions, regeneration aspects, and condition monitoring aspects.
Despite due care is taken by manufacturers and utility, yet there is a high probability for electrical discharges and local hotspots. The presence of electrical discharges and local hotspots for longer duration leads to high electrical discharge and thermal failures. The electrical faults are low and high-energy discharge electrical faults, which are treated as corona discharges, partial discharges, and breakdown. Thermal faults are the accumulation of heat over a local area and are treated as hotspots. Both electrical and thermal faults are originated because of various reasons including, dielectric defects, air bubbles, impurities, sharp edges, and any manufacturing defects. The presence of these faults hampers the integrity of the insulation system in various aspects and is detrimental to a transformer. Therefore, early detection of these faults is of high importance to condition monitoring engineers and maintenance planners. It is well established that, these faults are involved in the generation of several gasses that are dissolved in the oil. Therefore, the dissolved gas analysis is seen as one of the important conditions monitoring and diagnosis tools.
Numerous researchers have reported the gassing tendency of various dielectric liquids under different fault conditions. The tendency to gassing generation under partial discharges or low energy discharge electric faults in mineral oils and ester fluids were investigated and reported in [14][15][16]. It is reported that higher quantities of C 2 H 2 and H 2 are generated in natural esters and C 2 H 2 remains the key gas to indicate a defect or fault witnessed by electrical discharges. Also, the influence of high energy discharges on the gassing tendency of the dielectric fluids have been reported in [14,17,18]. Similar to the low energy discharge faults, H 2 and C 2 H 2 are found to be witnessed due to arcing fault conditions. However, a large amount of C 2 H 2 is generated as compared to the low energy discharge cases. The gassing tendency of ester fluids under thermal faulty condition was also investigated and are reported in [14,15,[18][19][20]. It is found that natural ester generates more gases at a higher temperature (>300 • C) and traces of C 2 H 6 are found increasing with increasing temperature. The key gasses in ester fluids under thermal faults are found to be CH 4 , C 2 H 4 , CO, and C 2 H 6 . However, there are different methods for simulating the hotspot at different temperatures; the gassing tendency of the dielectric fluid varies accordingly. The details are critically analyzed and summarized in [21]. It is to be mentioned that, the fault diagnosis approaches discussed above are limited to a single fault i.e., electrical or thermal fault alone. However, in the real transformer, there are possibilities for the electrical and thermal faults to occur simultaneously. Thus, in [22] a combination of electrical and thermal faults was studied. The authors reported the gassing tendency at different combinations of electrical and thermal faults while varying fault temperatures and fault voltages. The trending of fault gasses has been explicitly reported under different laboratory conditions. Electrical discharges, electrical arcing, and thermal faults in a transformer influence the degradation of the oil accompanied by faults gas generations. These faults also, expedite the degradation reactions in a transformer, which together with acids and heat accelerate deterioration of the insulation system. It is to be understood that the intensity of this degradation process and fault gasses are also attributable to the condition of the oil. In other words, the occurrence of the same fault with the dielectric fluid at two different conditions (age) will have different levels of consequences. Recently, authors from our research group have reported the influence of aging on the gassing tendency and oil's degradation under low energy electrical discharge faults for mineral oil and synthetic esters [23].
It was reported that the condition (age) of the oil has a significant influence on the concentration of the dissolved gases which affects the fault diagnostic conditions. Also, the influence of thermal aging on the gassing tendency of mineral oil and natural esters under the influence of thermal faults has been reported in [19]. A significant change in the dissolved gas concentration has been noticed with an increase in the thermal aging duration of the dielectric fluids. Hence, there is a need to study Energies 2020, 13, 3472 3 of 16 the gassing behavior of ester-based dielectric fluids (both natural and ester) under the influence of electrical and thermal fault conditions at different aging conditions and the influence of the faults on oil degradation. Therefore, the present research is aimed to understand the influence of aging or state of the dielectric liquid on gassing tendency under different operating faulty conditions. The authors have also focused the influence of different faults on the degradation of the dielectric fluids. In the present work a mineral oil, a synthetic ester, and a natural ester have been investigated under low electrical energy discharge, high electrical energy arcing, and hotspot faulty conditions. The degradation of the insulation liquids is monitored by adopting UV spectroscopy and turbidity measurements as per ASTM D6802 and ASTM D6181 standard test methods respectively. The dissolved gas analysis measurements are performed as per ASTM D3612.

Samples
For the present study, a mineral oil (MO), a natural ester (NE), and a synthetic ester (SE) have been adopted. The properties of the fresh unused oils considered for this study are tabulated in Table 1. Unused mineral oil, synthetic ester, and natural ester have been subjected to thermal aging according to a modified ASTM D1934 standard method at 115 • C for 2000 h. The thermal aging is performed with the copper catalyst (3 g/L) and cellulose Kraft papers (1:20). After the thermal degradation procedure, highly degraded oil/paper insulation is subjected to sonication for 30 s using Qsonica Q1375 Sonicator. This sonication allowed to prepare a gentle solution of degraded oil and paper. Further, the individual solutions of MO, NE, and SE are separately added to unused MO, NE, and SE respectively. The proportions of the degraded solution added to unused oils allowed controlling the acid values of the oils. This facilitated having oil samples with a specific level of acidity that refers to a bad class of the oil aging as per [24].
It is to be mentioned that attaining a predetermined value of acidity is not feasible with the traditional thermal aging process. Therefore, the authors prepared a degraded solution which is later used to obtain a specific value of the acidity by adding it in suitable proportions. Now, these aged oils (obtained with the desired acidity values) along with fresh unused oils are used for the present study. The details of the acidity measurements of the bad class are illustrated in Table 2. High energy discharge faults are stimulated by repeated breakdowns to produce arcing in the oil volume. The breakdown of the insulation fluid is performed in the laboratory as per ASTM D 877. To simulate the sustainability of the arcing fault, 100 breakdowns have been performed with 2 min of break between each breakdown. To understand the impact of high-energy discharge electrical faults, dissolved gas analysis is performed after 100 continuous breakdowns. To understand the influence on oil degradation, UV spectroscopy and turbidity measurements are reported before and after 100 breakdowns. The details of the breakdown test set up are presented in Figure 1a and the series of breakdowns for MO, NE, and SE for fresh and aged oils are shown in Figure 1b. In addition, the total average energy used for 100 breakdowns in various liquids is tabulated in Table 3.

High Energy Discharges (Arcing)
High energy discharge faults are stimulated by repeated breakdowns to produce arcing in the oil volume. The breakdown of the insulation fluid is performed in the laboratory as per ASTM D 877. To simulate the sustainability of the arcing fault, 100 breakdowns have been performed with 2 min of break between each breakdown. To understand the impact of high-energy discharge electrical faults, dissolved gas analysis is performed after 100 continuous breakdowns. To understand the influence on oil degradation, UV spectroscopy and turbidity measurements are reported before and after 100 breakdowns. The details of the breakdown test set up are presented in Figure 1a and the series of breakdowns for MO, NE, and SE for fresh and aged oils are shown in Figure 1b. In addition, the total average energy used for 100 breakdowns in various liquids is tabulated in Table 3.   It is observed that, breakdown of the dielectric liquids is increasing with aging and with the number of breakdowns. To understand this, it is to be recalled that water saturation limit of dielectric liquids increases with increase in aging [25]. Also, breakdown voltage is dependent on relative moisture content and hence breakdown voltage of the aged oil will be higher than that of the fresh oil at a given absolute moisture content. Under AC stress, a field-induced drift of particles, located in high field regions, can be entailed by a stress relieve leading to an improvement of the dielectric strength, compared to technically clean oil [26]. In the same way, breakdowns cause definite degradation; this will lead to an increase in the water saturation on average with breakdown times.

Low Energy Discharges (Partial Discharges)
Low energy discharges electrical faults are often called partial discharges, which could be internal discharge, surface discharge, and corona discharges. In the present work, a laboratory set up has been used to produce a surface discharge of 9 kV. A 9 kV electrical discharging activity is simulated for 5 h on the surface of the dielectric fluid. The free electrons are generated by a cylindrical copper electrode sealed in a 500 mL Erlenmeyer glass. The electrode is placed in the center of the discharge cell and suspended above the oil's surface. The distance between the central electrode and the surface of oil is approximately one inch. The details of the experimental set-up was reported by the authors from our research group in [23] and is shown in Figure 2a. The discharge is initiated after vacuuming the test cell and the changes in the pressure with discharging time are recorded to understand the gassing behavior of the insulating fluid. This method is useful in understanding the stability of the insulating fluids to gassing behavior. The pressure changes in the test cell with fresh and aged MO, NE, and SE as a function of discharging time are shown in Figure 2.
Energies 2020, 13, 3472 It is observed that, breakdown of the dielectric liquids is increasing with aging and with the number of breakdowns. To understand this, it is to be recalled that water saturation limit of dielectric liquids increases with increase in aging [25]. Also, breakdown voltage is dependent on relative moisture content and hence breakdown voltage of the aged oil will be higher than that of the fresh oil at a given absolute moisture content. Under AC stress, a field-induced drift of particles, located in high field regions, can be entailed by a stress relieve leading to an improvement of the dielectric strength, compared to technically clean oil [26]. In the same way, breakdowns cause definite degradation; this will lead to an increase in the water saturation on average with breakdown times.

Low Energy Discharges (Partial Discharges)
Low energy discharges electrical faults are often called partial discharges, which could be internal discharge, surface discharge, and corona discharges. In the present work, a laboratory set up has been used to produce a surface discharge of 9 kV. A 9 kV electrical discharging activity is simulated for 5 h on the surface of the dielectric fluid. The free electrons are generated by a cylindrical copper electrode sealed in a 500 mL Erlenmeyer glass. The electrode is placed in the center of the discharge cell and suspended above the oil's surface. The distance between the central electrode and the surface of oil is approximately one inch. The details of the experimental set-up was reported by the authors from our research group in [23] and is shown in Figure 2a. The discharge is initiated after vacuuming the test cell and the changes in the pressure with discharging time are recorded to understand the gassing behavior of the insulating fluid. This method is useful in understanding the stability of the insulating fluids to gassing behavior. The pressure changes in the test cell with fresh and aged MO, NE, and SE as a function of discharging time are shown in Figure 2.

Thermal Faults (Hotspot)
Thermal faults or hotspots simulation in MO, NE, and SE have been simulated by developing a laboratory-based test set up. The setup used for simulating thermal faults is presented in Figure 3a. The set up consists of a borosilicate-glass vessel and a Teflon cover with bushings to hold the heating element. Constantan was chosen for the heating element for its flat resistivity on a large scale of temperature. The input supply is connected to a variac that feeds two high current transformers

Thermal Faults (Hotspot)
Thermal faults or hotspots simulation in MO, NE, and SE have been simulated by developing a laboratory-based test set up. The setup used for simulating thermal faults is presented in Figure 3a. The set up consists of a borosilicate-glass vessel and a Teflon cover with bushings to hold the heating element. Constantan was chosen for the heating element for its flat resistivity on a large scale of temperature. The input supply is connected to a variac that feeds two high current transformers connected in parallel while the temperature of the wire was regulated by an ammeter and voltmeter connected between the heating element and the secondary. By measuring the voltage and current, the average temperature of the hotspot may be approximated. This is also validated by using a high-temperature resistant thermocouple. An aluminum funnel-shaped structure is placed just over the heating element. This is because the generated gas bubble enters the pipette through this funnel structure according to the Archimedes' principle where they can be measured. In each experiment, a quantity of 800 mL of dielectric fluid was introduced to fulfil the vessel and the pipette up to the reference value. The fluid samples under hot-spot tests are allowed to heat for periods of 30 min and the volume of gas generated is recorded. The temperature reached values close to 250 • C, which for the mineral is risky because of its flashpoint. For safety reasons, the current is zeroed until the temperature of the insulating fluid equals the room temperature (~24 • C). This test cycle of 30 min of heating and cooling down to room temperature have been repeated for 12 times. For every 30 min of the heating cycle, the volume of gas is recorded. The volumes of the gases generated as a function of the heating cycles are presented in Figure 3b.
Energies 2020, 13, 3472 6 of 15 the average temperature of the hotspot may be approximated. This is also validated by using a hightemperature resistant thermocouple. An aluminum funnel-shaped structure is placed just over the heating element. This is because the generated gas bubble enters the pipette through this funnel structure according to the Archimedes' principle where they can be measured. In each experiment, a quantity of 800 mL of dielectric fluid was introduced to fulfil the vessel and the pipette up to the reference value. The fluid samples under hot-spot tests are allowed to heat for periods of 30 min and the volume of gas generated is recorded. The temperature reached values close to 250 °C, which for the mineral is risky because of its flashpoint. For safety reasons, the current is zeroed until the temperature of the insulating fluid equals the room temperature (~24 °C). This test cycle of 30 min of heating and cooling down to room temperature have been repeated for 12 times. For every 30 min of the heating cycle, the volume of gas is recorded. The volumes of the gases generated as a function of the heating cycles are presented in Figure 3b.

Results and Discussions
The results of the degradation rate of the oil, before and after simulating electrical and thermal faults are discussed in this section. As mentioned earlier, the degradation of the insulating liquids has been understood by ultraviolet visible spectral curves, dissolved decay contents, and turbidity. The details of the fault gasses evolved because of the characteristic faults have been analyzed by dissolved gas analysis and application of Duval's diagnostic methods reported in the literature. To have a clear observation and to avoid direct comparison of various insulating oils, the authors have presented the results oil by oil with all types of faults. Also, it is difficult to obtain similar level of the fault intensity in order to compare fault by fault. Therefore, it is interesting to see the impact of the same type of faults in a dielectric fluid at different aging conditions (Fresh and Bad classes). The following notations presented in Table 4 are used for representing different cases of the insulating liquids.

Results and Discussions
The results of the degradation rate of the oil, before and after simulating electrical and thermal faults are discussed in this section. As mentioned earlier, the degradation of the insulating liquids has been understood by ultraviolet visible spectral curves, dissolved decay contents, and turbidity. The details of the fault gasses evolved because of the characteristic faults have been analyzed by dissolved gas analysis and application of Duval's diagnostic methods reported in the literature. To have a clear observation and to avoid direct comparison of various insulating oils, the authors have presented the results oil by oil with all types of faults. Also, it is difficult to obtain similar level of the fault intensity in order to compare fault by fault. Therefore, it is interesting to see the impact of the same type of faults in a dielectric fluid at different aging conditions (Fresh and Bad classes). The following notations presented in Table 4 are used for representing different cases of the insulating liquids.

Degradation of Insulating Fluids
In the present section, the degradation of fresh and aged oils while subjected to electrical and thermal faults is of concern. Fresh oils are technically clean and are free from decay particles and cellulose particles, and other aging byproducts. Whereas, aged oils contain, cellulose particles, dissolved decay products, acids, and other aging byproducts. Therefore, when a fresh oil and aged oil are subjected to different fault conditions, the undergoing chemical perspectives are expected to be different. Subsequently, the influence of these faults on the dielectric fluids will be different. Hence oil absorbance, dissolved decay products, and turbidity are measured for before and after subjecting the liquids to the fault conditions. The details have been presented in Figure 4.
In fresh oils, the changes that occur because of the application of thermal and electrical faults depend majorly on the thermal stability of the chemical bonds and the ionization potential respectively. In case of aged oils, along with the said factors, the concentration of colloidal and soluble decay particles plays a significant role in further degrading a dielectric fluid. Therefore, in the present research, the concentration of the dissolved decay products and turbidity measurements have been reported. From Figure 4, it is seen that aged oils have a high turbidity, decay content concentration which is obvious. It is to be noticed that, the application of fault conditions (both electrical and thermal) had a different impact for fresh and aged oils. In Figure 4a, from fresh mineral oil, the influence of partial discharges is high as compared to arcing and hotspot. Similarly, in the case of aged oils, the hotspot influence is superior to that of the partial discharges and arcing. However, it is seen that the absorbance of aged oil is reduced with arcing. This may be due to the splitting or de-agglomeration of the cellulose particles present in the oil which reduces the absorption to light. To further understand this, the chemical function group analysis is to be understood. The absorbance and turbidity of the aged oil are increasing with partial discharges and thermal faults. From Figure 4b, it is noticed that, the influence of hotspot is higher than that of the partial discharges and arcing for fresh and aged synthetic esters. Also, the influence of partial discharges and arcing is too low in the case of fresh and aged fluid samples. One may understand that, the influence of electric faults is low when compared to thermal faults in the case of synthetic esters. However, it is to be assured that, the intensity of the fault nature is uniform in all three cases. Also, in Figure 4c, the influence of the hotspot is higher when compared to the other faults in the case of fresh natural esters. To understand the influence, faults of much higher intensities are to be evaluated. It is to be recalled that, thermal faults involve an increase in local temperature and promoting oxidation of the oil/paper insulation. These generated acids trigger a self-sustained process called hydrolysis which in turn degrades the solid insulation. But, in case of electrical faults, Energies 2020, 13, 3472 8 of 16 the influence is more associated to a generation of acids and free radicals which leads the degradation. However, both electrical and thermal faults are involved with the generation of dissolved gasses. The detailed analysis of dissolved gas analysis is presented in the next section.

Degradation of Insulating Fluids
In the present section, the degradation of fresh and aged oils while subjected to electrical and thermal faults is of concern. Fresh oils are technically clean and are free from decay particles and cellulose particles, and other aging byproducts. Whereas, aged oils contain, cellulose particles, dissolved decay products, acids, and other aging byproducts. Therefore, when a fresh oil and aged oil are subjected to different fault conditions, the undergoing chemical perspectives are expected to be different. Subsequently, the influence of these faults on the dielectric fluids will be different. Hence oil absorbance, dissolved decay products, and turbidity are measured for before and after subjecting the liquids to the fault conditions. The details have been presented in Figure 4.

Gassing Tendency of Insulating Fluids
The dissolved gas analysis for mineral oil and non-mineral oils are analyzed according to IEEE C57.104 and IEEE C57.155 respectively. As per of IEEE C57.104 and IEEE C57.155 the following are the observation on gas trending in mineral oil and non-mineral oils respectively for different fault conditions.
Arcing faults: With high energy discharges, traces of C e H 2 and C 2 H e increase in mineral oils and C 2 H 6 is generated in ester fluids. However, H 2 , CH 4 , C 2 H 4 , C 2 H 2 are witnessed in common with arcing faults and C 2 H 2 need proper monitoring to diagnose arcing failures.
Energies 2020, 13, 3472 9 of 16 Partial discharge faults: With low energy discharges in the initial stages CH 4 and small traces of C 2 H 2 increases and at an advanced stage; traces of C 2 H 2 and C 2 H 4 raises in mineral oils. In ester fluids, H 2 and C 2 H 2 are witnesses. However, severe traces of H 2 remains common in both dielectric liquids. Hence, H 2 and C 2 H 2 may be monitored for identifying low energy discharging activities.
Thermal faults: Local hotspots in mineral oil increases H 2 , CH 4 , C 2 H 4 , C 2 H 6 , and C 2 H 2 whereas in ester oil, H 2 , CH 4 and C 2 H 6 , and some C 2 H 4 increases.
However, electrical and thermal faults are mostly associated with cellulose insulation and degradation of cellulose is accompanied by the generation of CO and CO 2 gases. Hence, the above discussed faults are also mostly witnessed by CO and CO 2 gases for cellulose-based insulation systems. Also, the influence of aging on the gassing behavior is not addressed in the said standards. In the present work, dissolved gas analysis has been performed on fresh and aged mineral oil and ester fluids after subjecting to partial discharges, arcing, and thermal hotspot. Hence, the gases generated under the influence of different faults for a dielectric liquid with aging are discussed in this section. The details of the dissolved gases for different oils under different faults for mineral oil, natural, and synthetic esters are shown in the Figure 5. It is noticed that, the concentration of the majority of the gasses is higher in case of aged oil under the influence of thermal and electrical faults. For arcing faults, H2, CH4, and C2H2 are found to increase in all the dielectric fluids. However, O2 is reduced and significant change is not noticed in the traces of N2 for fresh and aged oils. Thus, H2, CH4, and C2H2 gas concentrations vary with the state condition of the oil under arcing faults for mineral oils and non-mineral oils. For partial discharge faults, a much less change in H2 and C2H2 is noticed for fresh and aged oils. This is because, cellulose decay particles act as a filtering and absorbing medium under low energy discharge fault conditions [27]. However, there is almost no change in CH4, C2H4, and C2H6 for fresh and aged mineral oil. For ester fluids, a noticeable change in the traces of CH4, C2H4, and C2H6 are noticed. Therefore, under partial discharging activity, CH4, C2H4, and C2H6 vary with the condition of the ester fluids and H2 will be generated in common for both the fluids according to the state of the oil. For thermal faults, there is a significant change in H2, CH4, C2H4, and C2H6 in mineral oil while in ester fluids, there is a small change in CH4, C2H4, and C2H6. This is because of the fault temperature which is not more than 400 It is noticed that, the concentration of the majority of the gasses is higher in case of aged oil under the influence of thermal and electrical faults. For arcing faults, H 2 , CH 4 , and C 2 H 2 are found to increase in all the dielectric fluids. However, O 2 is reduced and significant change is not noticed in the traces of N 2 for fresh and aged oils. Thus, H 2 , CH 4 , and C 2 H 2 gas concentrations vary with the state condition of the oil under arcing faults for mineral oils and non-mineral oils. For partial discharge faults, a much less change in H 2 and C 2 H 2 is noticed for fresh and aged oils. This is because, cellulose decay particles act as a filtering and absorbing medium under low energy discharge fault conditions [27].
However, there is almost no change in CH 4 , C 2 H 4 , and C 2 H 6 for fresh and aged mineral oil. For ester fluids, a noticeable change in the traces of CH 4 , C 2 H 4 , and C 2 H 6 are noticed. Therefore, under partial discharging activity, CH 4 , C 2 H 4 , and C 2 H 6 vary with the condition of the ester fluids and H 2 will be generated in common for both the fluids according to the state of the oil. For thermal faults, there is a significant change in H 2 , CH 4 , C 2 H 4 , and C 2 H 6 in mineral oil while in ester fluids, there is a small change in CH 4 , C 2 H 4 , and C 2 H 6 . This is because of the fault temperature which is not more than 400 • C and hence a significant change in C 2 H 6 and C 2 H 2 are not witnessed. Hence, under thermal faults with less temperatures, H 2 , CH 4 , C 2 H 4 , and C 2 H 6 gases will be generated in common for both the fluids according to the state of the oil. From Figure 5b, it is noticed that, the generation of combustible gasses increase with increasing degradation of the oil. However, the generation of combustible gasses are higher for partial discharging and hotspot activity. This is because, the degradation of dielectric fluids is accelerated by cellulose decay particles, and other conductive particles. Therefore, excessive amounts of CO and H 2 increases easily in any event of the fault, thus aiding to the increase in TDCG. The illustration of Duval's triangle and Duval's pentagon with the fault distribution locations are shown in Figure 6 for mineral oil, natural, and synthetic esters.  To further understand the influence of the condition of the dielectric liquid on gassing tendencies under different faults, Duval's triangle and Duval's pentagon methods for mineral oil and non-mineral oils have been applied for mineral oil and ester fluids respectively. Both fault diagnosing methods indicate that faults evolve with degradation duration. However, this phenomenal changing depends on the intensity of fault and the degradation rate of the insulation system. The insulation system degradation rate is highly dynamic and hence the gassing tendency of the dielectric fluid remains dynamic with operation time. Because as per Duval's methods, any disagreement or mismatching in the fault type using triangle and pentagon methods indicate the presence of more than one fault [28]. Such a measurement requires more investigations to diagnose the fault. To illustrate changes of fault gas, both diagnosing methods and authors' observations on fault matching have been summarized in Table 5. It is to be noticed that, there is a high disagreement between all partial discharge fault samples for all the dielectric fluids. It is to be recalled that, in the present study, a spark is discharged on the surface of the liquid applying 9 kV voltage for five hours. The stray gassing is noticed for fresh dielectric fluids in case of partial discharging activity of the present study. One may understand that even minor partial discharging activities with low energy and for short durations will lead to stray gassing and involves in local temperature raise. To further comment on this, a detailed study on the type of fault diagnosis for partial discharges with different energy levels and different types (surface discharge, internal discharge, and corona discharge) is required. For arcing faults, the fault diagnosis matched for fresh samples (mineral oil, and ester fluids). However, aged mineral oil does not match but the similarity may be accepted as they are electrical discharge faults with different intensities. This is due to the aging impact; it is to be recalled that cellulose decay particles have been manually added to get the required acidity values.
It is known that the degradation of cellulose is much higher in mineral oils as compared to ester fluids. Hence, these decay particles added to the oil, differently influence the oil with arcing when compared to the ester fluids. Hotspot simulation faults almost matched or partially matched for all samples. However, fresh natural ester confined to stray gassing while aged natural ester result in thermal fault in case of Duval's pentagon whereas aged oils confined to thermal faults.
It is to be mentioned that care is taken during the experimentation to avoid interaction of the liquid with external air. Albeit all the three setups are hermetically sealed, a neck headspace is allowed in the test cells. This neck headspace is facilitated for the safety purposes, oil expansion, and to easily accommodate the bubble pressure created. Thus, there is a scope for the fault gasses to escape in the gas phase too. The above results are inclusion of only the fault gas that is dissolved in the liquid phase. To understand the gases in the gas phase, Ostwald solubility coefficient (L) of different gasses presented in IEEE Std C57.104 are used. The details of the Ostwald coefficients, gas phase gasses, liquid phase gasses, and total gas generation are presented in the Table 6.
It is to be mentioned that, the details of Ostwald solubility coefficients in case of ester fluids are not yet reported in IEEE Std C57.155. Therefore, for illustration purposes the coefficients reported in IEEE Std C57.104 have been used in the present study. Further, it is already arguable for the use of single Ostwald coefficient even for mineral oil as the aging of oil affects the partial pressure of the gases. Therefore, it will be quite difficult to justify the use of a single Ostwald coefficient for both mineral and ester liquids. Hence, it is to be understood that, the changes/variations in the gassing tendency in different liquids at different aging conditions may also be attributable to these factors. The details of accuracy and repeatability of individual gases measurements are provided in Table 7.

Conclusions
The influence of the condition of the dielectric fluid on gassing has been reported for mineral oil, natural, and synthetic esters under electrical and thermal faults. The degradation under fault conditions has been reported using dissolved decay contents, absorbance, and turbidity. The dissolved gas analysis by the application of Duval's fault gas diagnostic methods are adopted to understand the gassing behavior of the dielectric fluids. It is inferred that, fault gases in a transformer evolve with respect to degradation of the dielectric fluid. As degradation of the fluid is highly dynamic with operating times, there is a need to understand the gassing tendency of an insulation liquid vis-à-vis aging markers. It is also found that, the concentration and type of dissolved gasses generated depend on the type of fault. In other words, the gassing tendency and fault gasses will not be the same for a similar type of fault at different fault durations and fault energy. The theoretical premise that aging by-products affects the gassing tendency and DGA is experimentally confirmed. The results reported in this study should be helpful in further improving the dissolved gas analysis of ester filed transformers. Dissolved gas analysis is a potential fault diagnosing tool having a large variance in the research results reported by various researchers across the globe. Thus, demanding need to emphasize the research on dissolved gas analysis of alternative dielectric fluids at different aging conditions, different fault intensities, and fault durations.