Insulation Residual Life Estimation Based on Transformer Condition Assessment Data

• Published in: Electrical Insulation Conference and Electrical Manufacturing Expo pp. 1 - 4
• Main Authors: Sbravati, Alan, Cheim, Luiz V., Frimpong, George K.
• Format: Conference Proceeding
• Language: English
• Published: IEEE 18.06.2023
• Subjects: Degradation; Life expectancy; Loading; Moisture; Oil insulation; paper drying; Paper Life; Temperature measurement; Temperature sensors; Water; Windings
• ISSN: 2576-6791
• DOI: 10.1109/EIC55835.2023.10177293

The residual life of a transformer is commonly associated with the estimated life of the insulation system, namely the solid insulation. While the temperature is the most relevant parameter for the calculation of the solid insulation life consumption, the influence of the water and oxygen content must be taken in consideration. The traditional paper life curves, commonly called "Arrhenius Curves" are traditionally focused on the hydrolytically catalyzed thermal degradation of the paper, assuming water content of less than 0.5% in the paper. Only recently, when the latest version of the IEC loading guide was published, charts presenting life expectancy curves for different levels of moisture and oxygen were introduced. The inclusion of such parameters into the calculation affects the largely, reducing the calculated lifespan of transformers to as much as two thirds of that predicted by the traditional curves. However, the challenge lays on how to estimate the actual water content in the paper in different regions of the windings, since there is currently no viable way to measure such values. This article presents an approach for continuously and iteratively estimate water content in different regions of the windings, based on the known initial conditions and the effective loading and ambient temperatures. The results from the model are then compared with the information obtained from online moisture and temperature sensors, and an artificial intelligence model is used to minimize the deviations. Thus, the proposed methodology allows assessing paper degradation rate more accurately. This information is of paramount relevance for asset management and for strategic loading decisions, allowing the inference of a failure risk associated both with the paper degradation and risk of bubbling, i e., microbubbles formation.