A Rapid Evaluation Method for Transformer Insulation Status

A method for rapid analysis of transformer insulation status indicated by moisture content in insulation paper was developed by microwave-assisted solvent extraction coupled with Fourier transform infrared spectroscopy (FTIR). The results show that acetonitrile is the best extraction solvent, and the second derivative based on the difference spectrum is obviously better than the difference spectrum. For the acetonitrile spectrum of extracting water, the sensitivity of hoh bending vibration 1680 cm−1 is higher than that of OH stretching vibration 3615 cm−1. For all test samples, the results of this method are basically consistent with the standard method and show excellent precision. The study shows that it is feasible to rapidly analyze the moisture content of transformer insulation paper by solvent extraction coupled with FTIR.


Introduction
The solid insulation of transformers is made of fibrous insulation material, in which moisture can accelerate the aging of oil-paper insulation and shorten the lifespan of transformers [1].Therefore, accurately measuring the moisture content of insulation paper (board) is of great significance for the safe operation of transformers [2].At present, the monitoring of moisture content in transformer insulation paper mainly adopts indirect evaluation and direct detection methods.The indirect evaluation method is mainly based on laboratory data, and there is still a lack of on-site data validation.The moisture balance curve method obtains the average moisture content of transformer insulation paper by obtaining the moisture distribution function in the paper [3,4].It can only be applied to insulation systems composed of new oil and paper, and cannot calculate the moisture content of transformer insulation paper during operation.The dielectric response law cannot distinguish the effects of moisture and aging on the dielectric response curve and is prone to interference during on-site application [5,6].The current standard method DL/T 449 [7] is a direct detection method calculated through the collection of insulation paper samples and laboratory analysis, which takes a long time, and the extraction process is greatly affected by factors such as temperature and humidity, stirring consistency, and paper dissolution.At the same time, the Karl Fischer test agent used for trace water detection is harmful to personnel and the environment.
In this study, we utilized microwave assisted extraction coupled FTIR technology for quickly analyzing the moisture content of transformer insulation paper as Figure 1: 1) pre-treatment: We take a certain quality of insulation paper sample and quickly cut it into pieces with scissors; 2) Water extraction [8]: We place the paper sample into a microwave digestion tank and add methanol as the extraction liquid.At the same time, we prepare a microwave digestion tank and add an equal volume of methanol as the blank.We set parameters for microwave-assisted extraction; 3) water content detection: Using blank extraction solution as the background spectrum, we obtain the infrared spectrum of the sample, and calculate it into the standard curve after differential spectrum processing; 4) Drying: we immerse and clean the sample in the microwave digestion tank with petroleum ether, put it into a dry weighing bottle with constant weight, and weigh it after vacuum drying; 5) Calculation: we substitute the water content of the extraction solution sample and the quality of the insulation paper into the formula to calculate the water content of the insulation paper.

Extractants
To meet the conditions of suitable solvents, it should first be able to be miscible with water and have sufficient extraction power (strong polarity) to ensure complete extraction of water.At the same time, the solubility of oil on the extracted insulation paper should be as low as possible to reduce spectral interference.According to this, four polar solvents, dimethyl sulfoxide, methanol, ethanol, and acetonitrile, which are insoluble with transformer oil, are compared.Paper samples of equal quality are taken, and the results are compared using the magnetic stirring extraction method recommended by DL/T449.After extracting the same sample using the above 4 polar solvents, the water content of the extraction solution was determined to be 760, 773, 580, and 752, respectively, ȝg/ml.The results indicate that except for ethanol, the other three solvents are suitable.Considering the strong toxicity of dimethyl sulfoxide and acetonitrile, methanol was selected as the extractant for the experiment.

Microwave Extraction Conditions
Temperature.Under a pressure of 30 atm, 0.3 g of paper sample saturated with humidity in the air was taken, and 25 ml of acetonitrile was added as the extractant.The extraction time was 15 minutes.The extraction temperatures were set at 100ႏ, 120ႏ, 130ႏ, and 140ႏ, respectively.The water content of the extraction solution was analyzed using the Karl Fischer method, and three parallel samples were made to explore the effect of extraction temperature on the water extraction rate in insulation paper.Simultaneously, we take paper samples of equal quality and compare the results using the magnetic stirring extraction method recommended by DL/T449.
The results indicate that due to the insufficient accuracy of microwave temperature control, the repeatability of the test results is greatly affected in Table 1; The short-term increase in extraction temperature increases the solubility of water in methanol.When the extraction temperature rises to 120ႏ, the extraction rate remains stable and is basically consistent with the standard method measurement values within the range of 120-130ႏ; Continuing to increase the extraction temperature to 140ႏ, the water content of the extraction solution rapidly increases.The reason may be that too high an extraction temperature may cause the decomposition of cellulose or excessive pressure in the digestion tank, resulting in the escape of a large amount of methanol vapor, leading to distorted extraction results.The extraction temperature selected for the experiment is 120ႏ Time.Under a pressure of 30 atm and an extraction temperature of 120ႏ, 0.3 g of paper sample saturated with humidity in the air was taken, and 25 ml of methanol was added as the extractant.The extraction time was set for 2.5, 5, 10, 15, 20, and 25 minutes, respectively.The water content of the extraction solution was analyzed using the Karl Fischer method to explore the effect of extraction time on the water extraction rate of the edge paper.Simultaneously, we take paper samples of equal quality and size and compare the results using the magnetic stirring extraction method recommended by DL/T449.The results showed in Figure 2 that with the extension of extraction time, the extraction rate of water gradually increased and reached its peak after 10 minutes.Continuing to extend the extraction time, the extraction rate remained unchanged.The extraction time selected for the experiment was 10 minutes.

Infrared Analysis
We prepare H2O-methanol Standard solutions with the water content of 0.2, 0.5, 1.0, 2.0, and 3.0 mg/ml, record the background spectrum S0 of methanol, place each quantitative standard sample in an infrared spectrometer for testing, record each spectrum as Sstd1, Sstd2, Sstd3, Sstd4, and Sstd5 in sequence, and obtain the difference spectrum series S through spectral subtraction operation, i.e., Sstd-S0_ dif.The characteristic peak of dif water is related to its moisture content, and the corresponding moisture standard curve is obtained [9].
Optical path.We install liquid cell intervals of 0.1, 0.5, and 1.0 mm respectively, and compare the absorption spectra of standard samples under different optical paths.The infrared absorption peaks of moisture are mainly concentrated in two regions, the OH stretching vibration with strong absorption intensity is 3616 cm -1 , and the HOH bending vibration with weak absorption intensity is 1631 cm -1 .According to Beer's law, the absorbance A is proportional to the product of the concentration c of the absorbing substance and the optical path length of the absorption cell.When using a 0.1mm sample cell, due to insufficient sensitivity, the absorption of different levels of standard samples does not meet Beer's law(Figure 3(a)(d)); Under a 0.5 mm sample cell, the changes in absorption spectra at 1680 cm -1 and 3615 cm -1 are basically consistent with the changes in standard sample concentration(Figure 3(b)(e)); Under the 1.0 mm sample cell, the absorption spectra of different levels of standard samples showed disordered changes(Figure 3(c)(f)), which may be due to the strong absorption of water by methanol at this cell length, causing the signal to be disproportionate to the concentration.A 0.5 mm optical path sample cell was selected for the experiment.(e) 3615 cm -1 -0.5 mm (f) 3615 cm -1 -1.0 mm Figure 3 Absorption spectra at 1680 cm -1 and 3615 cm -1 for different optical paths Spectral determination.The absorption peak area of different standard samples in a 0.5 mm optical path sample cell was used as the standard curve for determining the water content of the extraction solution.The linearity of the standard curve for O-H bending vibration at 1680 cm -1 reached 99.35% as Figure 4, while the standard curve at 3615 cm -1 was only 96.50% in Figure 5.When measuring the moisture content of the actual insulation paper extraction solution, it must be taken into account that a small amount of insulation oil should not interfere with the infrared spectrum at the characteristic peak of moisture quantification.Figure 6 shows the comparison between the insulation paper extraction solution sample and the standard sample spectrum, and the absorption peak at the 1680 cm -1 position on the spectrum is used to determine the moisture content without interference from actual sample.

Drying Process
Under various heating and drying methods under normal pressure, fiber quality will undergo a certain degree of loss when heated.Therefore, the vacuum drying method is adopted to isolate air under negative pressure [10], so that thermal decomposition does not occur during the removal of methanol and petroleum ether dissolved in the fibrous material, ensuring the accuracy of quality.Temperature.We move the paper sample of the same quality into the weighing bottle with constant weight and recorded mass, set the heating temperature of the vacuum oven to 30, 40, 60, 80ႏ, heat for 30 minutes under each temperature condition, cool it in the dryer to room temperature, weigh it, put it in the vacuum oven to continue heating for 15 minutes, weigh again, and repeat this step until the paper sample is constant weight.The results show that when the heating temperature reaches 60ႏ, the insulation paper sample can maintain constant weight within 30 minutes.
Time.We move the paper sample into the weighing bottle with constant weight and recorded mass, set the heating temperature of the vacuum oven to 60ႏ, heat it for 10, 15, 20, 25, 30 minutes, cool it in the dryer to room temperature, and then weigh it.The results showed that the quality of the insulation paper remained basically stable after a heating time of about 20 minutes.To ensure the drying quality, a drying time of 30 minutes was selected for the experiment.

Comparison
Samples #1, #2, and #3 were named after the 100th, 85 th , and 1st layers of insulation paper taken from the same solid insulation sleeve.We perform two tests on each sample, and the results are shown in the table.The experimental results show that the method can achieve good reproducibility for different transformer oil.Compared with the traditional method, the error is less than 10%, and the reliability of the results reached 0.45-3.85%instead of 8.5-23.5% in the past.

Conclusions
The rapid analysis technology of microwave assisted solvent extraction coupled with FTIR was studied to analyze the moisture content of insulation paper.The results showed that while the solvent extracted the moisture, the vast majority of transformer oil remained in the insulation paper, the matrix absorption interference was less, the analysis precision was high, and the accuracy was good.

Figure 1
Figure 1 Analysis process