A New Method to Detect Hydrocarbon Gases in mineral Oil Using NIR Spectroscopy and Multivariate Calibration

This paper presents methods to analyze gases generated from decomposition of the insulation oil in power transformer, based on Near Infrared Spectroscopy (NIR). The combination of NIR spectroscopy and multivariate calibration has allowed for obtaining information over the samples. The mineral oils are consisting primarily of hydrocarbons (methane, ethane, ethylene and acetylene) associated with faults in the transformer insulation system. Therefore, it becomes necessary to develop an alternative method to the classical method of analysis by DGA gas chromatography as the possibility of being directly applied on production lines. For the method and its application, a near infrared spectrophotometer - NIR, without moving parts and with an optical system based on acoustic optical tunable filter - AOTF, suitable for harsh environments was developed. The usage of the technique in quantitative and qualitative was validated by comparison with gas chromatography. A sample cell was adapted to the measuring cell of the spectrophotometer. In this cell was added the insulated oil and the measuring of gases in the gaseous phase starts from the principle that in a transformer, the key gases generated due to incipient faults migrate naturally from the oil to the transformer tank atmosphere until reach a chemical equilibrium. The measuring cell fixed optical path cell, with length of 540 mm and volume of 490 mL and glass window with 5 mm of the thickness. A sphere valve connects the sampling cell to the measuring cell, allowing the application of vacuum in the measuring cell and subsequent diffusion gases dissolved to the measure cell continuously. Two pumps are connected to the system: A reflux pump with flow ranging from 5 to 5000 mL/min, and a double stage vacuum pump of 29.7 in Hg and 56 L/min. All the system is connected to a nitrogen purge valve to ensure the cleaning of cells and gases circulating lines. The calibration set containing 12 gaseous mixes was prepared straight in the measuring cell by the injection of proper volume of each gaseous hydrocarbon through the sept. The concentration ranged from 0.5 to 2.0 % (v/v) for each one of the analytes (methane, ethane, ethylene, and acetylene). The results of the presented study indicate that Near Infrared Spectroscopy (NIR) along with statistical techniques are a tool to be used for the online monitoring of specific gases produced by the decomposition of mineral insulating oils. The developed NIR spectrophotometer has shown appropriated characteristics to be applied in a large range of concentration of the key gases, ranging from 400 to 20000 ppm (v/v) were successfully determined. This work presented very suitable alternative methods for the monitoring of power transformers, contributing to the state of the art for certain applications. The results indicate that Near Infrared Spectroscopy (NIR) with chemometrics appears as a tool to be used for monitoring gas species produced in the decomposition of mineral oils and represents a good model to be applied to the hydrocarbons present in real samples and for on-line monitoring of electrical equipments and other applications like petrochemical samples.