Development of gas chromatography for dissolved gas in oil based on nitrogen plasma discharge detection principle

The gas chromatography detection technology of dissolved gas in oil based on the nitrogen plasma discharge detection principle was studied. This technology detected the characteristic emission light emitted by the excited molecule after the ionization of the detected molecule motivated by the plasma generated by the nitrogen molecule under the action of a high-energy electromagnetic field and realized the quantitative detection of different molecules. Compared with the previous national standard commonly used hydrogen flame ionization detector (FID) and thermal conductivity detector (TCD) gas chromatography, it has a lower detection limit. The advanced chromatographic system with the original process realizes the detection of dissolved gas in transformer oil, which has good linearity and repeatability and can ensure the accuracy of measurement data. The analysis only used nitrogen and did not use hydrogen to prevent explosion hazards. Nitrogen is at least ten times cheaper than helium and is more economical than helium ion chromatography in consumables. The device can detect the small content of the characteristic gas in the transformer oil and the slight change of the content to prevent the occurrence of dangerous accidents.


Introduction
With the rapid development of the domestic power system and the progress of the ladder, UHV transmission circuit equipment devices throughout the country.The safe and stable operation of UHV transformer equipment is very critical to the power supply system.When UHV transformer equipment fails, it may lead to a factory, a park, or even a city blackout accident, with extremely serious consequences.
Transformer equipment is often filled with transformer oil, and the composition and content of dissolved gas in oil can feedback on the running state of the transformer in real time.The dissolved gas mainly includes hydrogen (H 2 ), oxygen (O 2 ), nitrogen (N 2 ), carbon monoxide (CO), carbon dioxide (CO 2 ), methane (CH 4 ), ethylene (C 2 H 4 ), ethane (C 2 H 6 ) and acetylene (C 2 H 2 ), as well as C 3 and C 3 + .Among them, H 2 , CO, CO 2 , CH 4 , C 2 H 4 , C 2 H 6 and C 2 H 2 are valuable dissolved gas components for judging the internal faults of oil-filled electrical appliances.Under the normal operation of electrical equipment, the dissolved gas in the oil mainly comes from the transformer oil itself and only a tiny amount of gas decomposed during the aging and deterioration of the solid insulation material [1].The amount of these gases will increase rapidly when overheating fault, discharge fault or moisture occurs inside the electrical equipment.Most of these gases dissolve in the oil, and a small portion rises to the surface of the oil.Transformer oil is a mixture of many hydrocarbon molecules of different molecular weights, containing CH 3 , CH 2 and CH chemical groups bonded together by C-C bonds [2].Electrical or thermal faults can break some C-H bonds and C-C bonds, accompanied by the formation of a few active hydrogen atoms and unstable hydrocarbon radicals such as CH 3 * , CH 2 * , CH * or C * (including many more complex forms) [3].These hydrogen atoms or radicals are rapidly recombined through complicated chemical reactions to form hydrogen and low molecular weight hydrocarbon gases, such as methane, ethylene, ethane, acetylene, etc. and may also produce carbon solid particles and hydrocarbon polymers (X-wax).The formation of carbon monoxide and carbon dioxide in transformer oil is related to insulating paper.When the fault just occurs, the formation of gas dissolved in the oil; When the fault energy is large, it may also accumulate into free gas [4,5].Therefore, regular measurement of the composition and content of gas dissolved in oil is of great significance for the early detection of latent faults in oil-filled power equipment [6].In order to avoid misjudgment of electrical equipment failure, the power industry standard DL/T 722-2014 puts forward clear requirements for the detection period before, during and during the operation of new equipment and the detection under special circumstances.At the same time, it proposes specific index requirements for the dissolved gas content in the equipment before and during operation [7].
At present, in the national standard, the separation of oil and gas in transformer oil usually obtains the gas to be detected by incomplete separation combined with headspace gas extraction and oscillation separation [8,9].For the detection of lower hydrocarbons, such as methane, ethylene, ethane, acetylene, etc., the hydrogen flame ionization detector is usually used for detection; for hydrogen, the thermal conductivity detector is used for detection; for oxides such as carbon monoxide and carbon dioxide, it is necessary to first convert them into flammable gases (converted by a Ni converter), and then use a hydrogen flame ionization detector for detection [10].The entire detection method uses two detectors and a chemical conversion device.The detection method is cumbersome, the operation is complex, and the number of detectors used is large.Huang et al. [11] built a theoretical model for this method and gave guidance for practical analysis.Yang et al. [12] studied the error of the chromatographic analysis method.Chen [13] conducted experiments on photoacoustic spectroscopy and Raman spectroscopy detectors, and found that photoacoustic spectroscopy detection required a high experimental environment, while the detection limit of Raman spectroscopy was above 0.2 μL/L [14], which could not meet the detection requirements of ultra-high voltage transformers.Han et al. [15] used a self-made alternating current pulse helium ionization detector to analyze the dissolved gas, which proved the applicability of the helium ion detector.Jalbert, J., Gilbert and Tetreault used helium ion detector gas chromatography and headspace degassing method to detect dissolved gas in transformer oil [16].It is proved that the detection limit is high and there is room for improvement.However, the helium ionization detector needs to use strategic gas helium.The plasma detector is currently used as a new type of detector, usually using helium as the carrier gas.According to the response mechanism of the plasma detector.We designed a plasma chromatography with nitrogen as the carrier gas for analysis, which solves the problem that helium is not easy to obtain, and achieves the same detection limit as the helium ion detector.Simplify the cumbersome process of detecting dissolved gas components in transformer oil, and take into account safety and practicality.

Detection system design
There are two kinds of effects when the sample passes through the plasma: the characteristic radiation of the related elements and the ionization of the sample components after atomization [17].Two different types of detectors have been developed: one is the plasma atomic emission spectrometer for gas chromatography, and the other is the plasma ionization detector for gas chromatography.The current commercial plasma emission spectroscopy detector requires high power consumption and helium usage and is equipped with a spectrometer with a CCD array detector [18].The plasma detector is designed with nitrogen as carrier gas (Figure1).In order to make the molecule absorb more energy for glow discharge, a method of reducing gas concentration is used to enhance the absorption energy of a single molecule.
For nitrogen as a carrier gas, the plasma power is low, and the detection limit cannot meet the detection technical indicators.Therefore, a low-pressure device for plasma detectors is designed.The molecular vacuum pump is used to buffer the vacuum environment.At the same time, the negative pressure can be adjusted and displayed in real time.The current was adjusted, and the 337 nm, 780 nm and 225 nm polaroid were used for filtering (Figure 2).

Chromatography gas path process design
The chromatographic process needs to analyze characteristic gases such as hydrogen (H 2 ), carbon monoxide (CO), carbon dioxide (CO 2 ), methane (CH 4 ), ethylene (C 2 H 4 ), ethane (C 2 H 6 ), acetylene (C 2 H 2 ), etc.At the same time, it is necessary to prevent the pollution of the chromatographic gas path system by dissolved gases in oxygen (O 2 ), nitrogen (N 2 ) and oil above C 3 .The gas path analysis system process is designed for the above components, Figure 3.

Sampling process design
The gas sample dissolved in the oil was injected from the sample inlet 8, followed by No.1 port and No.10 port of the second switching valve 7, quantitative ring 9, No.3 port of the second switching valve 7, and finally from No.2 port of the second switching valve 7 to the sample outlet 10.

Analysis Process Design
After the second switching valve 7 is switched, the second carrier gas 11 carries the sample in the quantitative ring 9 through the first phase column 6; The samples in the first phase column 6 were preseparated and then flowed to the second phase column 17.The samples were pre-separated into three peaks in the second phase column 17: hydrogen oxygen nitrogen methane carbon monoxide peak, carbon dioxide ethylene acetylene peak, and propene propane acetylene peak.

Design principle of detecting hydrogen, oxygen, nitrogen, methane and carbon monoxide components
When the hydrogen oxygen nitrogen methane carbon monoxide peak completely flows from the second phase column 17 to the third phase column 19, it is separated into hydrogen oxygen nitrogen methane carbon monoxide again by the third phase column 19, and finally measured by the plasma detector 22.

Design principle of detecting carbon dioxide, ethylene, ethane and acetylene components
When the hydrogen oxygen nitrogen methane carbon monoxide peak completely flows from the second column 17 to the third column 19 and the carbon dioxide ethylene ethane acetylene just flows out from the second column 17, immediately switch the fourth switching valve 18, the first carrier gas 26 carries the ethylene ethane acetylene component pre-separated from the second column 17, and the carbon dioxide ethylene ethane acetylene component is separated again through the fourth column 25, and finally measured by the plasma detector 22.

Design principle of detecting propylene, propane and propyne components
When the carbon dioxide ethylene ethane acetylene peak completely flows from the second phase column 17 to the third phase column 19 and the phase, propane and phase just flow out from the second phase column 17, and immediately switch the third switching valve 15.The first carrier gas 26 carries the propylene, propane and propyne components pre-separated from the second phase column 17 and the propylene.Propane and propyne components are separated by the second phase column 17, and finally measured by the plasma detector 22.

Development of data processing system
The development of the data processing system is divided into analog signal access (signal amplification circuit development), software development, and other parts.The commercial plasma detector is equipped with a separate signal detector, but only the output of the original signal is not processed.The signal receiving and amplifying circuit is developed and designed.As shown in Figure 4 below, the current signal is very small (tens to hundreds of picoamperes) when accessing the Vin reverse input terminal from the detector signal output terminal.The signal reaches the reverse input terminal of the OPA through the resistance RP.Because the RP is 1, 000 euros and the RF is 10 times the RP, the signal after the amplifier circuit will be greatly improved.The multi-stage series connection uses a wider range of voltage-mode signals to access the ADC for digital quantization.The chromatographic operation software adopts 'software cooperates with hardware and application' to ensure that the modules in the system hardware have independence, versatility and cross-series reconfigurable combination and integration capabilities, and provides software that effectively cooperates with the rapid combination of hardware systems in the platform.The software design consists of five parts: one is the man-machine operation interface and system resource management design; the second is the module components and process control design of the instrument; third, data acquisition and processing program design; fourth, comparative analysis and calculation of data; fifth, instrument self-test program design.The entire software architecture is a top-down hierarchical structure, using a structured, standardized, functional, modular design method, with the phase column module, injection unit module, and detector module in the hardware to build a common software development and application platform.
The chromatographic operation software is divided into the bottom layer (lower computer), the middle layer (embedded touch screen and other hardware circuits) and the top layer (upper computer).The top layer is responsible for the man-machine operation interface, data analysis and processing of the comprehensive management platform.The middle layer is responsible for machine operation and control management.The bottom layer is the direct control layer to realize the basic control of the instrument, and the whole is to realize the information exchange and communication between different levels through Ethernet.
The overall design of the software adopts the modular design idea, including the data communication program, temperature control program, sampling control program, and three independent functional modules.
The software design of this system uses the median filtering algorithm.Median filtering is a nonlinear signal processing filtering method based on ordering statistics theory [19].Suppose that there is a time window of width 2n+1 (2n+1<<L) for a series of numbers L, denoted as m.Move over the sequence L to form the corresponding array {X i-n ... Xn... X i + n }, denoted by {Xi}.The approximation value of {Xi}is found to minimize the sum of the absolute error of {Xi}.The deduction is continued ①: ① In order to satisfy the above equality, the median value of the array {Xi} should be taken [20].The median filtering algorithm can effectively filter out too large or too small values, which not only avoids the introduction of the original value by the distortion of the sudden pulse interference signal, but also protects the distortion of the normal signal value due to processing.At the same time, random noise and balanced increase can be suppressed.The schematic diagram of the median filter program design process is shown in Figure 5.

Research and test of influencing factors of detector performance
The standard gas is produced by Dalian Date Gas Co., Ltd.The concentration of each component of dissolved gas in oil is shown in Table 1.The certificate number of the standard substance is 15-09680.The plasma detector (PED) first uses helium (99.999%) as the carrier gas and is connected using the built-in test software.After the plasma detector was connected to the software developed by Lansis Instrument (Shanghai)Co., Ltd, the peak was normal (Figure 6).Therefore, it is verified that the software can be used to calculate it, and the spectrum can be saved and can be integrated and other functions.Figure 6.Spectrum of the software developed by Lansis using helium as carrier gas

The influence of different vacuum degrees on the performance of PED detector
The conductivity of nitrogen is very easy to be broken down, resulting in weak discharge energy.The conductivity of nitrogen under vacuum decreases, which can increase the response signal of the detector (Figure 7).As can be seen from the above table (Table 2), the peak height of carbon dioxide (CO 2 ), ethylene (C 2 H 4 ), ethane (C 2 H 6 ) and acetylene (C 2 H 2 ) components gradually increase when the pressure value is from -65 kpa to -85 kpa.It can be seen from the above table (Table 3) that when the doping gas flow rate is adjusted from 5 ml/min to 45 ml/min, the peak heights of carbon dioxide (CO 2 ), ethylene (C 2 H 4 ), ethane (C 2 H 6 ), and acetylene (C 2 H 2 ) components gradually increase.When the doping gas flow rate is 25ml / min, the peak height of carbon dioxide (CO 2 ), ethylene (C 2 H 4 ), ethane (C 2 H 6 ) and acetylene (C 2 H 2 ) components reaches the highest value.At this time, when the doping gas flow rate is increased, the peak height of carbon dioxide (CO 2 ), ethylene (C 2 H 4 ), ethane (C 2 H 6 ) and acetylene (C 2 H 2 ) components gradually decrease.

Influence of gain on signals
Gain can be adjusted arbitrarily, and the value is multiple.However, when Gain increases, some signals will be distorted.When the optimal gain is Pre-Amp gain 3, the chromatographic peak of the dissolved gas component is the highest.
Table 4 shows that when the gain is adjusted from 1*1 to 4*2, the peak heights of carbon dioxide (CO 2 ), ethylene (C 2 H 4 ), ethane (C 2 H 6 ) and acetylene (C 2 H 2 ) components gradually increase.When the gain is 3*3, carbon dioxide (CO 2 ), ethylene (C 2 H 4 ) and ethane (C 2 H 6 ) reach the highest values of 356 mV, 412 mV and 473 mV, respectively.The peak height of acetylene (C 2 H 2 ) reaches the highest value of 161 mV when the gain is 4*1.

Comparison of the new instrument and national standard chromatography
The GB chromatography uses three detectors, a thermal conductivity detector, and two hydrogen flame ionization detectors.The thermal conductivity detector detects hydrogen.One hydrogen flame ionization detector detects carbon monoxide and carbon dioxide converted to methane by the reformer, and the other detects methane, ethane, ethylene, acetylene, and other hydrocarbons.This method has many detectors and requires hydrogen as auxiliary gas.Once it is leaked, it is easy to explode and the risk is high.The new instrument uses a plasma discharge detector that uses safe, inexpensive nitrogen gas as a carrier gas to detect all components to be measured.

The precision of the new instrument
. Typical spectra of LX3200 standard gases Table 5 and Table 6 are consecutive repeatability of the retention time and peak area of the standard gas determined by LX-3200, respectively (Figure 8).The relative standard deviation (RSD) of retention time is within 1%, and the relative standard deviation (RSD) of peak area is within 3%.It can be proved that the precision of the developed instrument meets the requirements of analysis and testing.

The detection limit of the new instrument for seven gases
Table 7 is the detection limit of the national standard, the theoretical detection limit of the common three-detection principle, and the theoretical detection limit of the LX-3200 plasma detector.

Linearity of the new instrument for seven gases
Figure 9 is the standard curve spectrum of hydrogen after the dilution of different multiples of standard gases.The standard curve is made at the lowest concentration of 2.05×10 -6 V/V to 1027×10 -6 V/V, and the linear correlation R 2 is 0.9979.Figure 10 shows the standard curve spectrum of carbon dioxide after the dilution of standard gases at different multiples.The standard curve is made at the lowest concentration of 6.2×10 -6 V/V to 3112×10 -6 V/V, and the linear correlation R 2 is 0.9973.Figure 11 shows the spectrum of the standard curve of ethylene after different dilutions of the standard gas.The standard curve is made at the lowest concentration of 0.2×10 -6 V/V to 97.9×10 -6 V/V, and the linear correlation R 2 is 0.9961.Figure 12 shows the standard curve spectra of ethane after the dilution of standard gases at different multiples.The standard curve is made at the lowest concentration of 0.2×10 -6 V/V to 102×10 -6 V/V, and the linear correlation R 2 is 0.9954.Figure 13 shows the standard curve spectrum of acetylene after the dilution of standard gas of different multiples.The standard curve is made at the lowest concentration of 0.1×10 -6 V/V to 51.3×10 - 6 V/V, and the linear correlation R 2 is 0.9954.At the low concentration of 0.1 ppm, the peak still reached 20700 uV, showing a good peak at a low concentration.Figure 14 shows the standard curve spectrum of methane after the dilution of different multiples of standard gas.The standard curve is made at the lowest concentration of 0.2×10 -6 V/V to 98.2×10 -6 V/V, and the linear correlation R 2 is 0.994.Figure 15 shows the standard curve spectrum of carbon monoxide after the dilution of different multiples of standard gas.The standard curve is made at the lowest concentration of 7.02×10 -6 V/V to 702×10 -6 V/V, and the linear correlation R 2 is 0.9988.Through the linear experiment, it is found that the plasma detector still has good linearity and low sensitivity when the standard gas is diluted 500 times.

Conclusions
At present, the dissolved gas analysis process stipulated by the national standard requires the use of hydrogen dangerous gas, while the helium ion chromatography process studied before requires the use of strategic gas helium.Our country is a poor helium country.Most of the helium is imported.Helium is also relatively expensive.To solve the above problems, we developed a new instrument using nitrogen as a carrier gas plasma discharge detector with the new process that can fully realize the detection of dissolved gas in transformer oil, has good linearity and repeatability, and can ensure the accuracy of measurement data.Compared with the previous national standard detector, it has a lower detection limit, and only nitrogen is used to avoid the explosion risk caused by dangerous hydrogen.Nitrogen is an autonomous gas, which is not only cheaper than helium, but also can detect the small content and slight change of characteristic gas in transformer, so as to prevent dangerous accidents in advance.

Figure 1 .
Figure 1.Schematic diagram of low voltage device of plasma detector Illustrate with cuts: 1-solenoid valve normally open, 2-exhaust port, 3-vacuum port, 4-filter, 5vacuum pump, 6-needle valve, 7-check valve, 8-electronic pressure display panel, 9-negative pressure gauge, 10-switch, 11-power adapter, 12-power plug.The current commercial plasma emission spectroscopy detector requires high power consumption and helium usage and is equipped with a spectrometer with a CCD array detector[18].The plasma detector is designed with nitrogen as carrier gas (Figure1).In order to make the molecule absorb more energy for glow discharge, a method of reducing gas concentration is used to enhance the absorption energy of a single molecule.For nitrogen as a carrier gas, the plasma power is low, and the detection limit cannot meet the detection technical indicators.Therefore, a low-pressure device for plasma detectors is designed.The molecular vacuum pump is used to buffer the vacuum environment.At the same time, the negative pressure can be adjusted and displayed in real time.The current was adjusted, and the 337 nm, 780 nm and 225 nm polaroid were used for filtering (Figure2).

Figure 2 .
Figure 2. Schematic diagram of plasma detector structure Note: OWM is the optical wavelength module and CPM is the chromatographic processing module.The geometry of the plasma cell is for conceptual purposes only.The size in the figure does not represent the actual design size.

Figure 4 .
Figure 4. Schematic diagram of plasma signal amplifying circuit

Figure 5 .
Figure 5. Schematic diagram of program design flow of median filtering algorithm

Figure 7 .
Figure 7. Response signal chromatogram of plasma detector at low conductivity

Figure 9 .
Figure 9.Standard curve of H 2

Figure 11 .
Figure 11.Standard curve of C 2 H 4

Figure 12 .
Figure 12.Standard curve of C 2 H 6

Figure 13 .
Figure 13.Standard curve of C 2 H 2

Figure 15 .
Figure 15.Standard curve of CO

Table 1 .
The component of standard gas

Table 2 .
Response of common impurities under different vacuum

Table 3 .
The influence of different doping gas flow rates on signals . The effect of gain on the signal

Table 7 .
Contrast The noise of the data is 4 μV; ② The noise of the data is 20 μV