Abstract
This paper summarizes the results from a research project that has explored how dynamic loads, more specifically those associated with high-powered ferry charging, affects the temperature, reliability, and estimated lifetime of distribution transformers. A large amount of data has been collected from relevant transformers. Using this data, long synthetic time series were generated using machine learning methods were generated. Machine learning was used to ensure realistic seasonal and stochastic variations in the data. These long time series were then given to a thermal model and used to calculate equally long series of hotspot temperature data. Finally, this series of hotspot temperatures were used as input for an ageing model. This was done to estimate the ageing rates of the transformers. The findings were that, based on the model used, the hotspot temperatures were low compared to thresholds set by the IEC loading guide. This was despite loads being over nominal load for short periods of time. The ageing models on the other hand, indicated that the rapid load variation could negatively affect the lifespan of the transformer insulation