One-Step Electrical Insulating Oil Regeneration on Electret PVDF/BaTiO3 Composite Nanofibers

Insulating oil is a pivotal component of power transformers, but it suffers from aging byproducts during service operation. The aging byproducts from the degradation of oil insulation tend to induce insulation failure, which poses a significant threat to the security of the power grid. Therefore, the regeneration of insulating oil is required to prolong the useful life of insulating oil and hence be of economic and ecological interests. Typical in-use oil regeneration routes employ multi-step procedures. In this work, a one-step regeneration method using a PVDF/BaTiO3 composite membrane is proposed. BaTiO3 endows the composite membrane with improved hydrophobicity and an electret state. The regeneration performance of the PVDF/BaTiO3 nanofiber membrane was assessed by considering the acid value, moisture content, dielectric loss factor tan δ, and the AC breakdown voltage of the refreshed oil. The test results showed that the filtration efficiencies toward formic acid and moisture were up to 77.5% and 60.6%, respectively. Moreover, the dielectric loss factor tan δ of the refreshed oil decreased evidently at a power frequency, and the AC breakdown voltage rose from 23.7 kV to 38.9 kV. This suggests that the PVDF/BaTiO3 composite membrane may be employed efficiently, and it minimizes aging byproducts via the one-step filtration.


Introduction
A power transformer is one of the key pieces of equipment in electrical substation systems [1][2][3][4]. As the key component of oil-immersed power transformers, insulating oil can extend the service life of power transformers by keeping the windings cool and suppressing arc generation [5][6][7]. Approximately 0.4 million tons of insulating oil are used worldwide in power equipment [8]. However, falling victim to the long-term impact of the thermal field and electromagnetic field, insulating oil is consistently undergoing oxidation that provokes byproducts such as carboxylic acid, ketone, aldehyde, alcohol, moisture, etc. [9,10] These byproducts elevate the acid value, moisture content, dielectric loss, and electrical conductivity of insulating oil, which can lead to permanent insulation failure and severe damage to power transformers and electrical substations.
The acidic substance may cause a significant dielectric loss of insulating oil, an increase in the corrosion rate of coils, and form sludges on the windings in power transformers. This reduces the heat transfer capability of the oil [11]. The decomposition products also attack the paper insulation of the transformer windings, and the paper loses its mechanical and dielectric properties [12]. Meanwhile, the increase in moisture content will predominantly reduce the breakdown voltage of insulating oil. High temperature cracking, aging, oxidative cracking, and hydrolysis of insulating paper within the power transformer promote the production of organic acids. Subsequently, these aging products will continue to accelerate the aging process. Instead of replacing insulating oil, the regeneration of insulating oil is a promising method that prolongs the transformer life and reduces carbon emissions [13][14][15].

Material
The PVDF powder (FR904, Mw = 680,000) was supplied by Shanghai 3F, China. The Ndimethyl formamide (DMF) with a purity better than 99.5% was considered as a solvent and was purchased from Chengdu Chron Chemicals, China. They were both used as received without further purification. Polypropylene (PP) nonwoven membranes of 250 µm in thickness with a weight density of 30 g/m 2 were provided by Zhejiang Zhaohui, China. The 40 nm barium titanium trioxide (BaTiO 3 ) nanoparticles with a purity better than 99.99% were provided by Aladdin Chemicals, China.

Nanofiber Preparation
In order to optimize the electrospinning parameters, the PVDF in a DMF solution with a concentration of 12 wt% was prepared. The BaTiO 3 was dispersed first in DMF by sonication, and then both solutes, BaTiO 3 and PVDF, were stirred further by adding the proper amount of PVDF to the solution to achieve the weight ratio of BaTiO 3 /PVDF, varying from 0 to 15 wt% by steps of 2.5 wt%. All the prepared solutions were subjected to degassing in vacuum at room temperature (ca. 25 • C).
The setup for electrospinning is shown in Figure 1. The electrospinning was performed at 35 • C and a relative humidity of 35%. The distance between the stainless-steel needle tip (radius of 0.8 mm) and the grounded collector was 20 cm. The voltage on the tip was set to 20 kV by a DC power supply, while a digital syringe pump was used to control the propulsive rate. The fabrication process during electrospinning lasted about 3 h and differed due to the variation in BaTiO 3 contents in the different test groups. The higher the BaTiO 3 content, the longer the time needed for electrospinning. h and differed due to the variation in BaTiO3 contents in the different test groups. The higher the BaTiO3 content, the longer the time needed for electrospinning. .

Characterizations
The morphology of the nanofibers was observed by a field-emission scanning electron microscope (FESEM, MIR3TM, Tescan, Brno, Czech). The SEM was equipped with a high-accuracy energy-dispersive spectroscope (EDS, GENESIS XM, EDAX, Osaka, Japan), which was feasible to examine the distribution of nanoparticles within the PVDF matrix. The hydrophobicity of the nanofibers was quantified by measuring the water contact angles.
In order to identify the charge stability of the nanofibers, corona charging was performed by a tri-electrode system, as in Figure 2. During the 3 min corona charging, the voltages at the needle and the grid were set to 7 and 2 kV, respectively. An electrostatic voltmeter (542A, Trek, Lockport, NY, USA) was utilized to monitor the surface potential decay (SPD) of the samples at 90 °C.

Characterizations
The morphology of the nanofibers was observed by a field-emission scanning electron microscope (FESEM, MIR3TM, Tescan, Brno, Czech). The SEM was equipped with a highaccuracy energy-dispersive spectroscope (EDS, GENESIS XM, EDAX, Osaka, Japan), which was feasible to examine the distribution of nanoparticles within the PVDF matrix. The hydrophobicity of the nanofibers was quantified by measuring the water contact angles.
In order to identify the charge stability of the nanofibers, corona charging was performed by a tri-electrode system, as in Figure 2. During the 3 min corona charging, the voltages at the needle and the grid were set to 7 and 2 kV, respectively. An electrostatic voltmeter (542A, Trek, Lockport, NY, USA) was utilized to monitor the surface potential decay (SPD) of the samples at 90 • C. .

Characterizations
The morphology of the nanofibers was observed by a field-emission scanning electron microscope (FESEM, MIR3TM, Tescan, Brno, Czech). The SEM was equipped with a high-accuracy energy-dispersive spectroscope (EDS, GENESIS XM, EDAX, Osaka, Japan), which was feasible to examine the distribution of nanoparticles within the PVDF matrix. The hydrophobicity of the nanofibers was quantified by measuring the water contact angles.
In order to identify the charge stability of the nanofibers, corona charging was performed by a tri-electrode system, as in Figure 2. During the 3 min corona charging, the voltages at the needle and the grid were set to 7 and 2 kV, respectively. An electrostatic voltmeter (542A, Trek, Lockport, NY, USA) was utilized to monitor the surface potential decay (SPD) of the samples at 90 °C.  A sand core funnel equipped with a vacuum pump was employed to examine the oil regeneration performance of the nanofibers via measurements of the moisture, acid value, dielectric dissipation factor, resistance, and breakdown voltage. To testify the validity of the During the preparation procedure of the formic acid test, the Karamay 25# mineral oil was first dried at 50 • C at 50~100 Pa for 48 h. For every 2 g of mineral oil, 1 µL of formic acid was added. The resultant was then shaken in an ultrasonic cleaner for 30 min, to achieve homogeneous dispersion. The acid value of the manufactured mineral oil was 0.551 mg KOH/g. Figure 3a,b show the scanning electron microscope (SEM) images spun from PVDF solutions with either 6% or 12% concentrations. There are numerous bead strings in Figure 3a, which are likely due to the low solution concentration. Upon such a case, the "liquid fibers" were not able to be sufficiently solidified while flying from the energized tip to the grounded collector, which should result in the so-called "wet fibers". Because of the surface tension force, the "wet-fibers" tended to shrink and to form a ball-shaped morphology, i.e., the resultant bead strings in Figure 3a. The higher PVDF concentration turned the situation from numerous to none, as evidenced by the well-formed fibers in Figure 3b. It is indicated that the fibers, upon a proper concentration, e.g., 12%, are feasible to obtain "dry fibers" on the collector. An image-analyzing program was used to quantify the diameter distribution of the nanofibers, which is shown in Figure 3b. Figure 3c provides the diameter-dependent frequency distribution of the nanofibers. It is seen that more than 50% of nanofibers had diameters ranging from 130 to 170 nm, which indicates a well-narrowed distribution and is expected to realize controllable fabrication. A sand core funnel equipped with a vacuum pump was employed to examine the oil regeneration performance of the nanofibers via measurements of the moisture, acid value, dielectric dissipation factor, resistance, and breakdown voltage. To testify the validity of the filtration performance tests, at least six circle-shaped membrane samples with diameters of 2 cm were applied in each set of experiments.

Morphology of Nanofibers
During the preparation procedure of the formic acid test, the Karamay 25# mineral oil was first dried at 50 °C at 50~100 Pa for 48 h. For every 2 g of mineral oil, 1 μL of formic acid was added. The resultant was then shaken in an ultrasonic cleaner for 30 min, to achieve homogeneous dispersion. The acid value of the manufactured mineral oil was 0.551 mg KOH/g. Figure 3a,b show the scanning electron microscope (SEM) images spun from PVDF solutions with either 6% or 12% concentrations. There are numerous bead strings in Figure  3a, which are likely due to the low solution concentration. Upon such a case, the "liquid fibers" were not able to be sufficiently solidified while flying from the energized tip to the grounded collector, which should result in the so-called "wet fibers". Because of the surface tension force, the "wet-fibers" tended to shrink and to form a ball-shaped morphology, i.e., the resultant bead strings in Figure 3a. The higher PVDF concentration turned the situation from numerous to none, as evidenced by the well-formed fibers in Figure 3b. It is indicated that the fibers, upon a proper concentration, e.g., 12%, are feasible to obtain "dry fibers" on the collector. An image-analyzing program was used to quantify the diameter distribution of the nanofibers, which is shown in Figure 3b. Figure 3c provides the diameter-dependent frequency distribution of the nanofibers. It is seen that more than 50% of nanofibers had diameters ranging from 130 to 170 nm, which indicates a wellnarrowed distribution and is expected to realize controllable fabrication.  The majority of fiber diameters were in the range between 130 and 180 nm, with an average size of 150 nm. The coefficient value (C.V.) of the fiber diameter was kept at merely 0.25, suggesting that fiber diameter forms a homogenous distribution. In order to obtain membranes with expectable insulating oil regeneration properties, it can be concluded that a 12 wt% PVDF fabrication membrane is the most suitable condition.

Morphology of Nanofibers
In order to recognize the influence brought by the addition of BaTiO3, SEM analysis techniques identical to those applied in Figure 3 were utilized for Figure 4 as well. The results are shown in Table 1 and Figure 4. The majority of fiber diameters were in the range between 130 and 180 nm, with an average size of 150 nm. The coefficient value (C.V.) of the fiber diameter was kept at merely 0.25, suggesting that fiber diameter forms a homogenous distribution. In order to obtain membranes with expectable insulating oil regeneration properties, it can be concluded that a 12 wt% PVDF fabrication membrane is the most suitable condition.
In order to recognize the influence brought by the addition of BaTiO 3 , SEM analysis techniques identical to those applied in Figure 3 were utilized for Figure 4 as well. The results are shown in Table 1 and Figure 4.    Figure 4 and Table 1 listed a number of significant diameters of the nanofibers fabricated after the addition of BaTiO3 in the PVDF electro-spinning preparation solution. As BaTiO3 content increased, the average nanofiber diameter displayed a tendency to rise when BaTiO3 content was under 10 wt% and gradually dropped when BaTiO3 was higher. When the BaTiO3 concentration was kept at 10 wt%, the nanofibers reached a series of dimensions with an average pore size of 0.81 μm and an average fiber diameter of 174 nm.
Furthermore, the average nanofiber diameter from the group with 10 wt% BaTiO3 reached a homogenous distribution where the C.V. was narrowed down to ca. 0.26. These fiber dimensions are considered the most suitable for fabricating an effective membrane.
Finally, typical nanofibers with a proper concentration of 10 wt% are shown in the following SEM in Figure 4. It is recognized that the addition of 10 wt% BaTiO3 did not disturb the characteristics of the fiber condition from pure 12 wt% PVDF. Therefore, given all the fiber dimensions, it is safe to conclude that nanofibers prepared under a 10 wt% BaTiO3 concentration were in excellent condition.

Charge Stability of PVDF/BaTiO3 Nanofibers
Considering the intrinsic polar property of PVDF and BaTiO3 and the possible poling during electro-spinning, the obtained composite nanofibers were expected to contain both space charges and oriented dipole charges. Space charges are introduced during electrospinning, and dipole charges are possibly formed due to the electric field generated by space charges. Therefore, we used surface potential decay (SPD) to identify the charge stability of the composite nanofibers. Figure 5 includes the surface potential decay curves of composite nanofibers with BaTiO3 contents varying from 0 to 10 wt%. The surface potentials (SPs) of nanofibers, regardless of BaTiO3 content, all demonstrated relatively fast paced decay within the first 2 h. Subsequently, during the latter 10 h, the nanofibers with different BaTiO3 contents began to show evident differences in the SPD rates. Distinctive from pure PVDF fiber, the   Figure 4 and Table 1 listed a number of significant diameters of the nanofibers fabricated after the addition of BaTiO 3 in the PVDF electro-spinning preparation solution. As BaTiO 3 content increased, the average nanofiber diameter displayed a tendency to rise when BaTiO 3 content was under 10 wt% and gradually dropped when BaTiO 3 was higher. When the BaTiO 3 concentration was kept at 10 wt%, the nanofibers reached a series of dimensions with an average pore size of 0.81 µm and an average fiber diameter of 174 nm.
Furthermore, the average nanofiber diameter from the group with 10 wt% BaTiO 3 reached a homogenous distribution where the C.V. was narrowed down to ca. 0.26. These fiber dimensions are considered the most suitable for fabricating an effective membrane.
Finally, typical nanofibers with a proper concentration of 10 wt% are shown in the following SEM in Figure 4. It is recognized that the addition of 10 wt% BaTiO 3 did not disturb the characteristics of the fiber condition from pure 12 wt% PVDF. Therefore, given all the fiber dimensions, it is safe to conclude that nanofibers prepared under a 10 wt% BaTiO 3 concentration were in excellent condition.

Charge Stability of PVDF/BaTiO 3 Nanofibers
Considering the intrinsic polar property of PVDF and BaTiO 3 and the possible poling during electro-spinning, the obtained composite nanofibers were expected to contain both space charges and oriented dipole charges. Space charges are introduced during electrospinning, and dipole charges are possibly formed due to the electric field generated by space charges. Therefore, we used surface potential decay (SPD) to identify the charge stability of the composite nanofibers. Figure 5 includes the surface potential decay curves of composite nanofibers with BaTiO 3 contents varying from 0 to 10 wt%. The surface potentials (SPs) of nanofibers, regardless of BaTiO 3 content, all demonstrated relatively fast paced decay within the first 2 h. Subsequently, during the latter 10 h, the nanofibers with different BaTiO 3 contents began to show evident differences in the SPD rates. Distinctive from pure PVDF fiber, the SPs of nanofibers, regardless of their BaTiO 3 contents, all displayed reduced decay rates that were evidenced by low SP value drops during the same time period, indicating that Polymers 2022, 14, 2631 6 of 13 the addition of BaTiO 3 was beneficial to the improvement of nanofiber charge stability. When the concentration of BaTiO 3 was higher or lower than 7.5 wt%, the decay rate of the surface potential was inversely or directly proportional to the doping concentration of BaTiO 3 , respectively. SPs of nanofibers, regardless of their BaTiO3 contents, all displayed reduced decay rates that were evidenced by low SP value drops during the same time period, indicating that the addition of BaTiO3 was beneficial to the improvement of nanofiber charge stability. When the concentration of BaTiO3 was higher or lower than 7.5 wt%, the decay rate of the surface potential was inversely or directly proportional to the doping concentration of BaTiO3, respectively. Normalized surface potential Since the Vs-i value remained at a higher degree during the same amount of time, there is reason to believe that the addition of BaTiO3 creates considerable interference, which is beneficial to the formation of more energy traps. Subsequently, excellent charge stability is achieved.
The decay of surface potentials for samples with 7.5 wt% and 10 wt% concentrations were comparatively slower than the other samples. Nevertheless, in the case where fibers shared the same thickness, larger surface potentials guaranteed a stronger and more stabilized electric field. Given that 10 wt% possessed the highest initial surface potential voltage, we can conclude that the optimal concentration of BaTiO3 for nanofiber charge stability is 10 wt%.

Hydrophobicity of PVDF/BaTiO3 Nanofiber Membrane
As moisture increases, the enhanced space charge accumulation and electric field distortion may pose a threat to the insulating properties of oil-paper insulation [20]. Therefore, the moisture content within the insulating oil should be confined. It is universally acknowledged that moisture removal is closely associated with the hydrophobicity of the filtration surface. Therefore, in order to achieve high moisture removal efficiency with membranes made out of PVDF/BaTiO3 nanofibers, their hydrophobicity ought to be probed.
The water contact angles (WCA) of nanofibers with different BaTiO3 concentrations were calculated by an optical contact angle measuring instrument, as previously mentioned in Section 2.3. The relationship between the WCA and the BaTiO3 doping concentration is shown in Figure 6: Since the Vs-i value remained at a higher degree during the same amount of time, there is reason to believe that the addition of BaTiO 3 creates considerable interference, which is beneficial to the formation of more energy traps. Subsequently, excellent charge stability is achieved.
The decay of surface potentials for samples with 7.5 wt% and 10 wt% concentrations were comparatively slower than the other samples. Nevertheless, in the case where fibers shared the same thickness, larger surface potentials guaranteed a stronger and more stabilized electric field. Given that 10 wt% possessed the highest initial surface potential voltage, we can conclude that the optimal concentration of BaTiO 3 for nanofiber charge stability is 10 wt%.

Hydrophobicity of PVDF/BaTiO 3 Nanofiber Membrane
As moisture increases, the enhanced space charge accumulation and electric field distortion may pose a threat to the insulating properties of oil-paper insulation [20]. Therefore, the moisture content within the insulating oil should be confined. It is universally acknowledged that moisture removal is closely associated with the hydrophobicity of the filtration surface. Therefore, in order to achieve high moisture removal efficiency with membranes made out of PVDF/BaTiO 3 nanofibers, their hydrophobicity ought to be probed.
The water contact angles (WCA) of nanofibers with different BaTiO 3 concentrations were calculated by an optical contact angle measuring instrument, as previously mentioned in Section 2.3. The relationship between the WCA and the BaTiO 3 doping concentration is shown in Figure 6: The water contact angle first increased with increasing doping concentratio eventually reached a maximum value of 149° at 10 wt%. The WCA started to de with further increases in the BaTiO3 doping concentration. The addition of BaTi creases the roughness of PVDF fibers and the air content within the grooves be rough fiber surfaces. Subsequently, the contact area between water and the mem surface is reduced, as the contact area between air and the membrane surface is incr Subsequently, the WCA of PVDF/BaTiO3 nanofibers is increased.
Such nanofibers with excellent hydrophobicity could be used in the process of efficiency insulating oil regeneration. When flowing insulating oil is subject PVDF/BaTiO3 composite membranes, it would be particularly difficult for moist penetrate through the layers within the membrane thanks to the high WCA. There is possible for moisture to be efficiently removed from flowing insulating oil.

A. Performance on Formic Acid Filtration
The mixture of various acids within the oil can be traced back to hydrolysis carbonate chain reaction during the aging process of insulating oil. Nevertheless, t hardly any chance for an accurate replica covering the infinite possibilities of aging lating. Therefore, in order to tap the acid removal potential of PVDF/BaTiO3 com membranes, formic acid was chosen as a substitute for acidic byproducts since it hol smallest particle dimension among all acids. Insulating oil with formic acid was pre as mentioned in 2.3. The acid values of differing test groups are listed in Figure 7. The water contact angle first increased with increasing doping concentration and eventually reached a maximum value of 149 • at 10 wt%. The WCA started to decrease with further increases in the BaTiO 3 doping concentration. The addition of BaTiO 3 increases the roughness of PVDF fibers and the air content within the grooves between rough fiber surfaces. Subsequently, the contact area between water and the membrane surface is reduced, as the contact area between air and the membrane surface is increased. Subsequently, the WCA of PVDF/BaTiO 3 nanofibers is increased.
Such nanofibers with excellent hydrophobicity could be used in the process of highefficiency insulating oil regeneration. When flowing insulating oil is subjected to PVDF/BaTiO 3 composite membranes, it would be particularly difficult for moisture to penetrate through the layers within the membrane thanks to the high WCA. Therefore, it is possible for moisture to be efficiently removed from flowing insulating oil.

A. Performance on Formic Acid Filtration
The mixture of various acids within the oil can be traced back to hydrolysis and a carbonate chain reaction during the aging process of insulating oil. Nevertheless, there is hardly any chance for an accurate replica covering the infinite possibilities of aging insulating. Therefore, in order to tap the acid removal potential of PVDF/BaTiO 3 composite membranes, formic acid was chosen as a substitute for acidic byproducts since it holds the smallest particle dimension among all acids. Insulating oil with formic acid was prepared as mentioned in 2.3. The acid values of differing test groups are listed in Figure 7. The removal efficiency of formic acid increased with the increase in the BaTiO centration. The acid value of the mineral oil mixture after filtration was inversely p tional to the doping concentration of BaTiO3, reaching a minimum of 0.124 mg KO 10 wt%.
As mentioned in Section 2.3., the initial acid value of prepared insulating oil i mg/L. Given the acid value measured after a one-time regeneration with a 1 PVDF/BaTiO3 composite membrane, the calculated results show that the formic a moval efficiency under such a condition approached 77.5%, which is the highes among all the samples. Once the doping concentration exceeded 10 wt%, the acid tended to rise as BaTiO3 doping concentration increased, which means that form removal efficiency was lowered.
The high acid filtration efficiency of 10 wt% BaTiO3 membrane is evidenced lowest acid value shown in Figure 7, suggesting that even formic acid can be sm removed with merely a one-time regeneration. Such progress is likely attributed homogenously distributed fiber diameters and the pore size of PVDF/BaTiO3 mem

B. Acid Value after Filtration
After comparing all test results from Figure 8 with the original acid value be regeneration (0.241 mg KOH/g), there is reason to believe that PVDF membranes ferent doping concentrations all possess the ability to reduce the acid content of a sulating oil, supporting the theory that PVDF/BaTiO3 membranes hold the poten filter all sorts of byproduct acids. Additionally, the excellent filtration behavior of t doping concentration membrane was spotted. The acid value of the insulating oil a first-time filtration was 0.11 mg KOH/g, and the filtration efficiency was 48.6%, whi 25.2% more efficient than the conventional PVDF membrane. The removal efficiency of formic acid increased with the increase in the BaTiO 3 concentration. The acid value of the mineral oil mixture after filtration was inversely proportional to the doping concentration of BaTiO 3 , reaching a minimum of 0.124 mg KOH/g at 10 wt%.
As mentioned in Section 2.3., the initial acid value of prepared insulating oil is 0.551 mg/L. Given the acid value measured after a one-time regeneration with a 10 wt% PVDF/BaTiO 3 composite membrane, the calculated results show that the formic acid removal efficiency under such a condition approached 77.5%, which is the highest value among all the samples. Once the doping concentration exceeded 10 wt%, the acid value tended to rise as BaTiO 3 doping concentration increased, which means that formic acid removal efficiency was lowered.
The high acid filtration efficiency of 10 wt% BaTiO 3 membrane is evidenced by the lowest acid value shown in Figure 7, suggesting that even formic acid can be smoothly removed with merely a one-time regeneration. Such progress is likely attributed to the homogenously distributed fiber diameters and the pore size of PVDF/BaTiO 3 membranes.

B. Acid Value after Filtration
After comparing all test results from Figure 8 with the original acid value before oil regeneration (0.241 mg KOH/g), there is reason to believe that PVDF membranes of different doping concentrations all possess the ability to reduce the acid content of aged insulating oil, supporting the theory that PVDF/BaTiO 3 membranes hold the potential to filter all sorts of byproduct acids. Additionally, the excellent filtration behavior of the 10% doping concentration membrane was spotted. The acid value of the insulating oil after the first-time filtration was 0.11 mg KOH/g, and the filtration efficiency was 48.6%, which was 25.2% more efficient than the conventional PVDF membrane. Since acids other than formic acid possess larger sizes on the molecule lev harder for acids to penetrate through pores on the filtration membrane. Thus, the co acid filtration efficiency of the PVDF/BaTiO3 composite membrane after a one-time eration is expected to be higher than 77.5%, which was the removal efficiency of acid.
Based on our results, we concluded that external particles in the aging insulat not only involve acidic colloid but also solid particles, most of which are larger than acid. Therefore, it is reasonable to speculate that the removal efficiency of our regen method towards acidic byproducts could be even higher than 77.5%.

C. Moisture content within insulating oil
The moisture content after oil regeneration acts as a rule to study the remov ciency of moisture. Figure 8 includes the moisture content in the insulating oil after eration with composite nanofibers with BaTiO3 contents varying from 0 to 10 wt% Given that the actual moisture content within aging insulating oil is 40.1 mg test results from Figure 9 pointed out that the PVDF/BaTiO3 membrane commits contribution to the significant drop in moisture content. The moisture content wit sulating oil reached a minimum of 15.8 mg/L after the regeneration with 10 wt% c tration, displaying a filtration efficiency at 60.6% and elevating the filtration efficie 3.2%. The quality of the insulating oil after the first-time filtration was merely 5% from meeting the general standards. Since acids other than formic acid possess larger sizes on the molecule level, it is harder for acids to penetrate through pores on the filtration membrane. Thus, the common acid filtration efficiency of the PVDF/BaTiO 3 composite membrane after a one-time regeneration is expected to be higher than 77.5%, which was the removal efficiency of formic acid.
Based on our results, we concluded that external particles in the aging insulating oil not only involve acidic colloid but also solid particles, most of which are larger than formic acid. Therefore, it is reasonable to speculate that the removal efficiency of our regeneration method towards acidic byproducts could be even higher than 77.5%.

C. Moisture content within insulating oil
The moisture content after oil regeneration acts as a rule to study the removal efficiency of moisture. Figure 8 includes the moisture content in the insulating oil after regeneration with composite nanofibers with BaTiO 3 contents varying from 0 to 10 wt%.
Given that the actual moisture content within aging insulating oil is 40.1 mg/L, the test results from Figure 9 pointed out that the PVDF/BaTiO 3 membrane commits a great contribution to the significant drop in moisture content. The moisture content within insulating oil reached a minimum of 15.8 mg/L after the regeneration with 10 wt% concentration, displaying a filtration efficiency at 60.6% and elevating the filtration efficiency by 3.2%. The quality of the insulating oil after the first-time filtration was merely 5% away from meeting the general standards.

D. Dielectric loss factor after filtration
The actual dielectric loss factor tan δ of aged insulating oil is 1.439% at 90 °C and 50 Hz. By comparison, the dielectric loss factor tan δ of insulating oil after the regeneration of nanofiber membranes with varying BaTiO3 concentrations from 0 to 15 wt% all displayed obvious decreasing in dielectric factor tan δ. When the BaTiO3 concentration within PVDF was kept at 10 wt%, dielectric loss factor tan δ was lowered to 0.958%, improving the dielectric loss factor by 33.4% and 19.9%, compared to aged insulating oil before and after filtration by a pure PVDF membrane, respectively. This quality of insulating oil outperformed the general standard, which is tan δ ≤ 2%, to be specific.
As is known to all, the dielectric loss factor is closely associated with conducting loss and polarization loss within the oil. More external particles contained in aged insulating oil will lead to a higher conductive carrier concentration and an enhanced conductive rate. Subsequently, a high conductivity leads to a high dielectric loss. Therefore, it can be concluded that the low dielectric loss observed in Figure 10 suggests that the PVDF/BaTiO3 membrane filtration efficiency on external particles is satisfactory.

D. Dielectric loss factor after filtration
The actual dielectric loss factor tan δ of aged insulating oil is 1.439% at 90 • C and 50 Hz. By comparison, the dielectric loss factor tan δ of insulating oil after the regeneration of nanofiber membranes with varying BaTiO 3 concentrations from 0 to 15 wt% all displayed obvious decreasing in dielectric factor tan δ. When the BaTiO 3 concentration within PVDF was kept at 10 wt%, dielectric loss factor tan δ was lowered to 0.958%, improving the dielectric loss factor by 33.4% and 19.9%, compared to aged insulating oil before and after filtration by a pure PVDF membrane, respectively. This quality of insulating oil outperformed the general standard, which is tan δ ≤ 2%, to be specific.
As is known to all, the dielectric loss factor is closely associated with conducting loss and polarization loss within the oil. More external particles contained in aged insulating oil will lead to a higher conductive carrier concentration and an enhanced conductive rate. Subsequently, a high conductivity leads to a high dielectric loss. Therefore, it can be concluded that the low dielectric loss observed in Figure 10 suggests that the PVDF/BaTiO 3 membrane filtration efficiency on external particles is satisfactory.

D. Dielectric loss factor after filtration
The actual dielectric loss factor tan δ of aged insulating oil is 1.439% at 90 °C and 50 Hz. By comparison, the dielectric loss factor tan δ of insulating oil after the regeneration of nanofiber membranes with varying BaTiO3 concentrations from 0 to 15 wt% all displayed obvious decreasing in dielectric factor tan δ. When the BaTiO3 concentration within PVDF was kept at 10 wt%, dielectric loss factor tan δ was lowered to 0.958%, improving the dielectric loss factor by 33.4% and 19.9%, compared to aged insulating oil before and after filtration by a pure PVDF membrane, respectively. This quality of insulating oil outperformed the general standard, which is tan δ ≤ 2%, to be specific.
As is known to all, the dielectric loss factor is closely associated with conducting loss and polarization loss within the oil. More external particles contained in aged insulating oil will lead to a higher conductive carrier concentration and an enhanced conductive rate. Subsequently, a high conductivity leads to a high dielectric loss. Therefore, it can be concluded that the low dielectric loss observed in Figure 10 suggests that the PVDF/BaTiO3 membrane filtration efficiency on external particles is satisfactory.  E. Breakdown Voltage after filtration Table 2 listed the sample oils for breakdown voltage test. The average initial breakdown voltages of samples A B were 23.7 kV and 50.1 kV, respectively, as shown in Figure 11. Samples D to I displayed higher breakdown voltages than those of samples B and C. Compared with sample B, sample G demonstrated the highest breakdown voltage of 38.9 kV, increasing it by 64.1%, while sample C merely elevated the breakdown voltage by 25.7%. This suggests that one-time filtration with PVDF/BaTiO 3 composite membranes partially recovered the insulating properties within the oil and possessed a higher filtration efficiency than one-time filtration with a pure PVDF membrane. This improvement outperformed the general standard for the relevant breakdown voltage (30 kV). Overall, it is reasonable to conclude that a one-step regeneration with PVDF/BaTiO 3 nanofibers is feasible to achieve higher efficiency insulating oil regeneration. E. Breakdown Voltage after filtration Table 2 listed the sample oils for breakdown voltage test. The average initial breakdown voltages of samples A B were 23.7 kV and 50.1 kV, respectively, as shown in Figure  11. Samples D to I displayed higher breakdown voltages than those of samples B and C. Compared with sample B, sample G demonstrated the highest breakdown voltage of 38.9 kV, increasing it by 64.1%, while sample C merely elevated the breakdown voltage by 25.7%. This suggests that one-time filtration with PVDF/BaTiO3 composite membranes partially recovered the insulating properties within the oil and possessed a higher filtration efficiency than one-time filtration with a pure PVDF membrane. This improvement outperformed the general standard for the relevant breakdown voltage (30 kV). Overall, it is reasonable to conclude that a one-step regeneration with PVDF/BaTiO3 nanofibers is feasible to achieve higher efficiency insulating oil regeneration.

Mechanism of High-Efficiency Regeneration of Insulating Oil
It is obvious to say that nanofibers have a significant advantage in removing fine particles from a fluid due to their excellent mechanical blocking effect, which has been evidenced by numerous studies. Thanks to the high porosity observed in the high-definition SEM figures (e.g., Figure 3b), the improved mechanical blocking does not bring a pressure drop increase during filtration. In addition, the hydrophobicity of the PVDF/BaTiO 3 composite nanofibers is raised by involving BaTiO 3 , and it plays a significant effect on removing moisture from insulating oil. This should be sourced to increase the fiberroughness of BaTiO 3 despite its relatively high hydrophilicity. Further, the addition of BaTiO 3 tends to enhance the electric performance of PVDF, which should reasonably intensify the local electric field around the nanofibers and therefore endows the possibility to polarize the aging byproducts. This is justified by the distinctive acid removal capability, particularly the performance of clearing formic acid, which is the smallest organic acid, in insulating oil (cf. Figure 7). With regard to the acidic aging byproducts in insulating oil, typically large-sized acid molecules such as acetic acid and furfural, the nanofibers present with a higher capacity for acid removal (cf. Figure 8), which provides evidence on a microscopic level. Last but not least, the testing results from the circle-shaped membrane samples mentioned in Section 2.3. proved that our composite membrane samples can process 1500-2000 mL of insulation oil regeneration. This suggests that our membranes hold great potential in commercialization.

Conclusions
The PVDF/BaTiO 3 composite nanofibers are fabricated by optimized electro-spinning parameters. The addition of BaTiO 3 into PVDF improves hydrophobicity and electrical properties. The regeneration was able to lower the acid value of aged insulating oil from 0.241 to 0.11 mg KOH/g. Further, the moisture in the oil dropped from 17.3 to 15.8 mg/L, and the breakdown voltage recovered by 64.14% (i.e., from 23.7 to 38.9 kV). The PVDF/BaTiO 3 composite nanofibers are potential candidates to regenerate aged insulating oil.

Data Availability Statement:
The data that support the findings of this study are available from the corresponding author upon reasonable request.