Assessing Polymerization-Based Diagnostics for Transformer Insulation via Comparative Statistics

Power transformers are essential for grid stability and efficient energy transfer, but their reliability declines due to aging insulation systems made of paper and mineral oil. Monitoring techniques such as oil testing, dissolved gas analysis (DGA), and furan compound analysis help assess degradation, with the degree of polymerization (DP) serving as a key indicator of insulation health. This study evaluates five DP estimation methods, namely Chendong, Heisler & Banzer, Vaurchex, Pahlavanpour, and De Pablo, using six statistical metrics consisting of average, standard deviation, determination coefficient (DC), correlation coefficient (CC), t-test, and p-value. The Chendong method proved most robust, achieving DC = 0.677, CC = 0.878, and the lowest standard deviation (0.81), meeting all criteria. Heisler & Banzer followed with DC = 0.529 and CC = 0.878, though its higher deviation (1.04) affected consistency. Vaurchex and Pahlavanpour showed moderate performance (DC = 0.674 and 0.435) but failed to meet t-test and p-value thresholds. De Pablo ranked lowest (DC = 0.071), meeting only one criterion. By quantifying each method’s strengths and limitations, this paper offers a benchmarking framework to improve insulation diagnostics and guide maintenance decisions which ultimately enhance transformer reliability, asset management, and power system efficiency.