Data Driven Transformer Thermal Modelling for Identification of Thermal Issues

Abstract

Electricity demand is increasing because of global decarbonisation efforts to reduce emissions that have an impact on climate change. Climate change also impacts the operation of electrical networks due to increasing ambient temperatures and extreme weather conditions. It is therefore a challenging time for power system owners to operate their assets as efficiently as possible. Power transformers are key components in the transmission systems and expensive assets. Due to economic and technical aspects, there has been much research studied to maximize their life expectancy while maintaining the reliability and stability. The two key aims of this thesis are to develop accurate transformer thermal models, utilising a range of operational data, and to assess the thermal performance of aged units. A top-oil thermal model is developed using thermal-electrical analogy and heat transfer principles that captures thermal influence of prevailing winds and solar radiation. The key improvements of the proposed thermal model are calculating the heat transfer coefficient of the radiator on the air side using the heat transfer coefficient of combined forced and natural convection and including the solar radiation as an addition heat source. The proposed model is validated against operational measurements. The results are also compared with the predictions based on the IEEE-Annex G model. The proposed model is generally more accurate in all periods, especially windy and sunny periods as expected. Condition monitoring of power transformers, which are key components in the systems, is essential to identify incipient faults and avoid catastrophic failures. Machine learning algorithms, i.e. nonlinear autoregressive neural networks with external inputs and support vector machine for regression are used to capture dependency of the transformer temperature on loading and weather conditions for purpose of monitoring. These thermal models are trained using historical measurements. The results are validated against field measurements and it clearly demonstrated that the alternative algorithms surpass the IEEE Annex G thermal model. An incipient thermal fault identification algorithm is then proposed and successfully used to identify an issue using measurements taken in the field. This algorithm could be used to alert the operator and plan interventions accordingly.

More information

Identifiers

ORCID for Paul Lewin: orcid.org/0000-0002-3299-2556

Catalogue record

Export record