Confirmation of necessity and applicability of using various methods of interpretation of results of gas chromatographic analysis of power transformers

The method of analysis of gases dissolved in oil is one of the most informative methods of early detection of defects in power oil-filled transformers. Now, the decision on the state of the transformers is based on the method of interpretation of the results of the gas chromatographic (GC) according to the guideline document RD 153-34.0-46.302-00. At the same time, there are situations when this document does not provide accurate analysis results. Thus, it is proposed to use several methods of interpreting the results of the GC to obtain refined conclusions. The purpose of the study is to substantiate the need to use various methods of interpretation of the results of gas chromatographic analysis of the oil of power transformers. The following methods for dissolved gas analysis have been reviewed: Rogers Ratio Method, IEC 60599 Standard Method, Doernenburg Ratio Method, Duval Triangle Method, ETRA method, as well as the guideline document RD 153-34.0-46.302-00 method adopted in the Russian Federation. These methods are implemented in various power companies of the Russian Federation, such as PJSC “Rosseti MR”, PJSC “FGC UES” and PJSC “Mosenergo”. The article reveals the need to consider the totality of all available methods and techniques based on RD 153-34.0-46.302-00 and development of training samples. The scientific novelty and significance of the conducted research lies in the confirmation of the need to use a set of methods for interpreting the results of the GC. An algorithm for the complex application of the methods described in the article for interpreting the results of the GC and training samples has been formed. The obtained results allow us to consider the possibility to develop software for the complex application of the methods of interpretation of the results of the GC described in the article and the formation of training samples based on the developed algorithm.