Eskom has completed a multi-phase investigation into the use of methanol and ethanol as supplementary indicators of cellulose insulation degradation in power transformers, assessing whether alcohol analysis can add value alongside established condition monitoring tools such as dissolved gas analysis (DGA) and furan (2-FAL) testing. This study was presented by Hlengiwe Nzima at CIGRE Southern Africa 2025.
Kraft paper remains the primary solid insulation material in large power transformers with its mechanical strength and remaining life closely linked to the degree of polymerisation (DP). As the paper ages under thermal stress or fault conditions, it degrades and produces by-products including carbon monoxide, carbon dioxide, furans and alcohols. International studies have suggested that methanol may serve as an early stage indicator of cellulose depolymerisation while ethanol may be associated with high-temperature ageing or localised hotspots.
Eskom’s study evaluated whether headspace gas chromatography with flame ionisation detection (HS-GC-FID), implemented in accordance with ASTM D8086-20, could reliably detect methanol and ethanol at diagnostically useful concentrations, and whether meaningful trends could be established across its transformer fleet. Laboratory testing confirmed that the method could achieve clear separation and quantification of both alcohols down to low parts per billion levels.
The investigation included the analysis of approximately 250 in-service transformer oil samples conducted in several phases. Early results from randomly selected samples – largely taken from free-breathing units using metal containers – proved inconclusive, highlighting the sensitivity of alcohol measurements to sampling method and storage conditions. Subsequent phases focused on syringe-sampled oils from sealed transformers, including units with thermally upgraded paper, to reduce sampling-related uncertainty.
In the focused dataset, methanol and ethanol were detected in many transformers that still indicated “excellent” estimated DP values, suggesting potential value for early-stage trending. Ethanol concentrations broadly followed trends observed for carbon monoxide and carbon dioxide while methanol was generally present at lower levels. However, no consistent linear relationships could be established between alcohol concentrations and DP, transformer age or total dissolved combustible gas across the broader population.
Nzima concluded that methanol and ethanol analysis should be treated as supplementary diagnostic indicators rather than replacements for established techniques such as furan analysis and DGA. He recommends continued monitoring of sealed transformers over a longer period – at least two to three years – to determine whether alcohol trending can meaningfully enhance transformer health indices and long-term asset management decisions.
