Compact Modeling of Moisture-Driven Dielectric Behavior in Oil-Paper Insulation Under Environmental Stress

This work presents a compact, data-driven model to characterize the moisture-dependent dielectric behaviour of oil paper insulation under varying environmental and operational stresses. The model integrates dielectric response parameters loss tangent (tan δ), transfer function zeros, and polarization current to Predict Moisture content (%pm) through a nonlinear relationship. It captures the effects of thermal aging, trapped charge, temperature variation, and moisture diffusion across a wide frequency range. Experimental validation using both mineral and ester-based fluids demonstrates improved accuracy over conventional Polarization and Depolarization Current (PDC) and Cole-Cole Distribution Model (CDM) methods, especially in low-moisture conditions typical of in-service transformers. Compensation techniques address residual polarization and ambient noise, enhancing real-world diagnostic reliability. The framework offers a scalable, cost-effective tool for moisture estimation and predictive maintenance. While Frequency Domain Spectroscopy (FDS) provides noise immunity and easier shielding than PDC, field application faces challenges. Accurate FDS requires charge-free, thermally stable insulation conditions often unmet in practice due to residual dipole energy and temperature fluctuations. Reliable interpretation thus demands correction for non-idealities like thermal gradients and trapped charges to ensure accurate insulation assessment. Major Findings: This research establishes the study proposes of a compact model for accurate moisture estimation in oil-paper insulation, capturing effects of thermal aging and trapped charges. It achieves superior accuracy, especially in low-moisture scenarios. The model provides a scalable solution for insulation condition monitoring.