Transformers, one of the most important components of the power system, play a significant role in facilitating transfer of power to end users. It is necessary to continuously monitor and assess transformers in order to ensure reliability and availability of the power supply.

In the factory, each transformer is subjected to a standard impulse test after manufacturing. The main objective of factory testing of the transformer is to ensure, verify, and document that the transformer has no faults in the winding. The test is based on a visual comparison of the primary current at the reduced and full Basic Insulation Level (BIL). This dissertation proposes a new mathematical method: Short Time Fourier Transform (STFT), to improve the fault detection sensitivity in the transformer impulse test. The results show that the fault detection sensitivity, including sensitivity to detect minor faults, is enhanced significantly using STFT in the transformer impulse test.

When a transformer is transported from the factory to the substation, the windings can be displaced with possible deformation. A low voltage impulse test presents an identification of winding deformations or short circuits, such as Low Voltage Impulse (LVI) and Sweeping Frequency Response Analysis (SFRA) tests. Winding problems in transformers are evaluated by comparing Transfer Functions (TF) obtained from the transformer under different conditions. Any significant deviations from the fingerprint, the test performed at the factory, may indicate the incidence of faults on the windings, e.g. short circuits. To obtain better signature analysis in the transformer maintenance test, this dissertation suggests a periodic and low voltage test, using Random Pulse Sequence (RPS) and Pulse Sequence (PS) in the transfer function analysis. The outcomes prove that superior signature analysis is obtained, and the failure detection sensitivity is improved.

The novel methods have been developed based on a numerical simulation, using EMTP, performed on a 50MVA power transformer. The methods have been validated experimentally on a distribution transformer, 25kVA, and power transformers, 26MVA and 192MVA. A quantitative analysis of the transfer function method is described. This dissertation will advance the maintenance techniques of transformers, which can be adopted by utilities.