Comparing Two Topology Transformer Hysteresis Models with Power Transformer Measurements

Power transformer modelling and simulation requires detailed material and design information. However, this detailed information about a transformer is usually not entirely available. Therefore, we use a supplemental single-phase excitation setup (saturation test) to measure the transformer core hysteresis characteristic at elevate voltages. Only transformer parameters from the factory acceptance test and very few design data is used to set up the topology transformer models. In topology-correct models each section of the core is represented separately. Also the different off-core (leakage) flux paths can be included in the models. In this study we compare two transformer topology models in the electric domain with measurements from a 3-limb 2-winding, 50 kVA transformer in YNyn0 vector group. The first model uses the capacitance-permeance analogy to model the transformer. The second model uses the principle of duality of electric and magnetic circuits together with the Jiles-Atherton (JA) model for the hysteresis implementation. The hysteresis model parameters in both models are identified with the Nelder-Mead Simplex algorithm. We have found that both transformer topology models can reproduce the measured current waveforms during the no-load and saturation test with very good accuracy. The active and reactive power deviation between the capacitor-permeance model and the measurements are below 10 %, and for the JA model below 17 % for the no-load and the saturation test.