Abstract
Background: Distribution transformers play a vital role in maintaining the stability of
electrical networks, and their premature failure can result in significant power disruptions and
financial losses. In Emfuleni Municipality, the prevalence of non-linear loads has led to
increased harmonic distortions, accelerating transformer aging. Despite the critical nature of
this issue, the combined effects of harmonic distortions and thermal stress on transformer
degradation remain underexplored in practical municipal settings.
Objective: This study investigates the impact of harmonics and thermal stresses on the
performance and lifespan of a 40 MVA, 88/11 kV distribution transformer using advanced
predictive modelling techniques.
Method: Harmonic load data were captured via a Chauvin PEL103 power quality meter, and
a modified IEEE temperature rise model was employed to assess losses, temperature profiles,
and insulation degradation. Key performance metrics included eddy current and stray losses,
top-oil and hotspot temperatures, and aging acceleration factors. Two predictive models—a
linear regression model and an artificial neural network (ANN)—were developed and
benchmarked against IEEE and IEC standards.
Results: Results revealed that harmonic distortions increased transformer losses by over 20%,
raised critical temperatures by 15–20%, and shortened the transformer's operational lifespan
by approximately 20–25%. The ANN model demonstrated superior accuracy with a mean
absolute percentage error (MAPE) of 0.673%.
Conclusion: The findings underscore the urgent need for effective harmonic mitigation
strategies and highlight the ANN model's potential for accurately predicting transformer aging.
Future work should focus on real-time monitoring and hybrid AI-based systems to enhance
transformer reliability and extend service life in high-stress environments.