A Radially Laminated Core for Partial Core Transformers

There are a number of areas of research for transformers at present which are focussed on reducing the losses in transformer cores as well as the steel weight. One way to reduce core losses is to remove the core altogether making an air-core transformer which have been primarily used in pulse transformers and superconducting transformers However these designs have issues with large magnetising currents. Alternatively, the use of partial-core transformers (PCT), which have been developed as resonant test transformers , power transformers, and superconducting transformers, can be seen as a compromise between the transformer performance and a reduction in the use of materials. In a PCT there is a steel core down the centre of the windings, however the steel limbs and yoke of a traditional core have been removed as shown in Figure 1. As the remainder of the magnetic flux path is now through air instead of steel, there is a much higher reluctance path in a PCT than a full core transformer [5]. The University of Canterbury (UoC) has developed a number of partial-core transformers (PCTs) in an attempt to reduce the electrical losses of the PCTs. Towards the ends of the core in a PCT, the magnetic flux tends to spread outwards radially within the core. This increases the eddy current losses in the core compared to that seen in traditional full core design. Thus, PCT designs tend to have significantly higher eddy current losses per unit weight in the core compared with full core counterparts. This paper starts by providing a background into partial core transformers and their applications as well as a look into the performance of transformer steel. A radially laminated core is then designed and constructed to be compared with a traditional parallel stacked core for use in PCT. The two cores have been tested under open circuit conditions and their core losses computed from measured results. The core losses have also been compared to the modelled core loss as a test of the model accuracy