Abstract
Online condition assessment of surge arresters is
necessary for early detection of problems and subsequent replacement
of the device to avoid unplanned downtime and/or damage
to the equipment being protected. The most common technique
of monitoring and assessing the condition of the arrester is extraction
of the resistive component of the leakage current. There
are many drawbacks to this method arising from inaccuracies in
the model-based approach and deficiencies in the understanding
of the leakage current signal. This paper presents leakage current
signature analysis (LCSA) as a method of assessing the condition
of the surge arrester where the harmonics components of the
current signal, extracted using Bayesian spectrum estimation, are
used as indicators of degradation. An experimental methodology
comprising measurement of the leakage current of arresters
undergoing constant-stress accelerated degradation testing is
employed. Results indicate that a consistent increase in the
DC component of the leakage current signature in the case of
degrading samples is the most likely indicator of condition.