الفهرس 
IntroductionOnly 14 pages are availabe for public view
 |  | from | 132 |  |  |
| | | from | 132 | | |
Abstract
The insulation level of electrical transformer oil may deteriorate due to electrical and thermal faults, which may lead to transformer failure and system outage. This ensures the importance of detecting and predicting the fault type. In this regard, the first part of this thesis presents a more accurate diagnostic technique for transformer faults. Two different proposed techniques are introduced, namely, proposed techniques (IOFM) and (IOTM) with the aim of increasing the diagnostic accuracy of transformer oil faults using the dissolved gas analysis (DGA) methodology. The proposed technique (IOFM) was built using a combination of five methods. These five methods were integrated in a comprehensive manner, enabling to get more accurate diagnosis compared to individual methods. The proposed technique (IOFM) demonstrated its exceptional accuracy when compared to other existing methods, utilizing a comprehensive dataset of 360 instances with known actual faults. Building upon this success, a further advancement was achieved by expanding the dataset to 532 instances and introducing a new approach that focused solely on two methods for analyzing dissolved gases: the Artificial Neural Network (ANN) method and the Conditional Probability method.
As dissolved gases are successfully used to detect actual oil faults in the first part, these dissolved gases can be used to predict the condition of the oil. These can be used alone or with several tests. For this regard, the second part of this thesis presents a predictive maintenance, specifically targeting the evaluation of the oil health index of a power transformer (40 MVA - 66/11 kV) filled with NYTRO 10XN oil type. Initially, the evaluation is carried out based only on dissolved gas analysis. Subsequently, additional tests are incorporated in the evaluation, including acidity, moisture content, and breakdown voltage measurements. The results obtained from this evaluation clearly indicate that employing multiple diagnostic tests for the oil evaluation of power transformers yields higher accuracy compared to relying on DGA alone. Furthermore, the findings reveal that the utilization of multiple tests brings the assessment closest to the actual condition of a 20-year-old transformer with similar ratings and the same oil type. Finally, according to the result of the second part which ensure that oil assessment using more than one diagnosis test is more accurate than using only DGA and provides closer evaluation to the actual condition, the last part of the thesis apply this concept for a novel Gas-To-Liquid (GTL) oil in comparison to conventional mineral oils. Both oils are exposed to accelerated aging process. The study examines two types of oil: Diala S₂, representing the mineral oil, and Diala S₄, representing the GTL oil. The properties of both oil types were analyzed at various aging periods (3, 6, and 10 days) to simulate thermal stress experienced during normal service conditions. For each of the two types of different oils, the three precipitated samples with different aging, as well as the new sample, were subjected to electrical tests such as AC breakdown voltage, dielectric breakdown factor (DDF), moisture, acidity, density, viscosity, Ultraviolet-visible absorption (UV-Vis) and photoluminescence (PL) tests. On the other hand, the DGA method was used to measure dissolved gases of the two oil types as a chemical test after simulating abnormal conditions through thermal heating tests at approximately 300°C (thermal heating test” " ~ " ” 300°C) of the two types. The results confirmed that GTL oil type can maintain its electrical, chemical and physical properties compared to other mineral oil with aging. By calculation the oil health index using several tests, it proved that the condition of GTL oil is better than mineral oil with aging more than 10 years.