CONDITION ASSESSMENT OF HIGH VOLTAGE POWER TRANSFORMER INSULATION SYSTEM USING MULTIPARAMETERS

• Main Author: Azis Prasojo, Rahman
• Format: Dissertations
• Language: Indonesia
• Subjects: Teknik (Rekayasa, enjinering dan kegiatan berkaitan)
• Online Access: https://digilib.itb.ac.id/gdl/view/63043
• Institution: Institut Teknologi Bandung

ABSTRACT CONDITION ASSESSMENT OF HIGH VOLTAGE POWER TRANSFORMER INSULATION SYSTEM USING MULTI- PARAMETERS By Rahman Azis Prasojo NIM: 33218020 (Doctoral Program in Electrical Engineering and Informatics) In the electric power system, power transformers are vital equipment with a large number of units and expensive. Problems that occur in high-voltage transformers can cause significant material losses, both for utilities, as well as for customers. To ensure that the transformer is in good condition to operate, assessment of the current condition is crucial. The weak point of a transformer based on a survey conducted by CIGRE 964 is the winding. The predominant failure mode is the dielectric, with the main cause of failure being the aging of the oil-paper insulation system. Therefore, this study focuses more on determining the condition of high-voltage power transformers through the integrity of the oil-paper insulation system. The transformers are routinely observed, measured, and examined. The variety of test parameters, as well as the large number of transformers managed, presents its own difficulties. Of these many test parameters, Health Index is a method that is widely used in determining the overall condition of a transformer. This Health Index value can be used to sort transformers in a population based on the transformers that require attention first. The conventional Health Index approach requires that all data be available to obtain an accurate transformer condition. The unavailability of data in the transformer Health Index analysis is one of the limitations of this concept according to CIGRE 761. Due to the unavailability of some data, the values obtained may hide important problems. Several studies have tried to solve this problem by predicting unavailable data using other parameters. Despite having satisfactory prediction results, these studies have not proposed a comprehensive method that integrates the prediction results into the transformer Health Index assessment. In the early stages, the use of seven models to replace furfural data which is often not available were investigated. The Health Index of 200 transformers with complete data was calculated using the previous method, and compared with calculations using an alternative model. The imputation approach to predict the condition of transformer paper using Multiple Linear Regression (MLR) and ANFIS (Adaptive Neuro-Fuzzy Inference System) has a better fit than other approaches indicated by a higher correlation coefficient with a complete Health Index, each of 0.959 and 0,960. In the next stage, the Health Index structure of the insulation system for high- voltage power transformers has been developed. The structure developed is divided into 3 factors, namely oil quality (OQF), faults (FF), and paper condition (PCF). Determination of the weighting factor utilizes input from experts using the AHP (Analytic Hierarchy Process) method. A new approach in determining FF using Faults Severity has also been developed based on a combination of gas level, gas rate, and DGA interpretation of the Duval Pentagon Method (DPM), resulting in more reliable assessments so that more accurate decisions can be made. The proposed Faults Severity level is more selective and shows more sensitivity in some cases, due to the use of multi-criteria. To overcome the problem of unavailable data, several previous studies have proposed predictions of important parameters that are often not available. Parameters that is promising to make predictions are Interfacial Tension (IFT) and paper condition (Furan/Degree of Polymerization[DP]). The Health Index concept developed can still be implemented even though the available data is limited, by only including the weighting factor of the available parameters. However, only reporting Health Index scores based on incomplete data can lead to misunderstandings. Therefore, in the third stage of this study, a model for calculating the level of certainty (referred to as Certainty Level / CL) was proposed. The CL number is determined from the number and level of criticality of the parameters available to be reported along the Health Index results. The proposed CL can also be used to integrate the predicted results of unavailable parameters into the transformer HI. After that, evaluation of the impact of data unavailability on HI results is carried out. A Random Forest-based Interfacial Tension (IFT) prediction model is also proposed. The accuracy of the proposed model is up to 90.91%. The integration of the prediction result to HI and CL calculation is also proposed. By reporting the value of CL, asset managers can gain better judgment in determining appropriate actions related to maintenance of high-voltage power transformers. In the final stage of the research, it was proven that the Health Index method developed can assess the condition of the transformer insulation system, and has high suitability with the actual condition of the transformer. The results of the evaluation show that the proposed Health Index method is more suitable than the previous method, as evidenced by a comparison to the Health Index calculation of out-of-service transformers with the average of 43.6 HI value, compared to previous method resulting in 59.68. Further implementation has also been proposed to develop predictions of remaining life using HI values, typical HI decreasing rates, and replacement criteria. Based on the observed population, HI of 40 is set to be the replacement criterion, 1.8 HI value per year as typical decrease, and the operation age of the entire population is estimated to be 36 years. vi vii The novelty proposed from this dissertation research is the development of high voltage power transformer insulation system Health Index with some changes compared to the previous method, such as different scoring, weighting, and structure. A new Faults Severity assessment has also been proposed, based on gas concentration, gas increasing rate, and DPM interpretation results. Certainty Level is proposed to overcome the problem of unavailable data. When the transformer data is incomplete, HI can still be calculated but by also reporting the Certainty Level along with the HI value. Certainty Level that has been developed can also accommodate the prediction results of data not available in HI. The resulting model is expected to help professionals diagnose high-voltage power transformers, so that damage can be detected as early as possible. With good condition monitoring and diagnosis of high-voltage transformers, appropriate maintenance scenarios can be arranged so that the transformer population management strategy can be implemented properly. Keywords: health index, high voltage power transformer, condition assessment, oil- paper insulation system.