Optimization of FRA by an Improved Numerical Winding Model: Disk Space Variation

Frequency response analysis (FRA) is a well-established method used for condition assessment of transformer windings and has proven its sensitivity for detecting various mechanical and electrical faults. Although the FRA test procedures have been standardised, the interpretation of FRA results is still a challenge as it is limited to analysis of the experts in the field. Mainly, circuit models are proposed in the literature for supporting the interpretation of the transformer frequency response. However, these concentrated parameters models are limited to a certain frequency range, due to the difficulties in calculating parameters to build and solve a turn-to-turn model. Moreover, constant values for the parameters are employed in these models, while these parameters are frequency dependent. In contrast, this paper presents an improved numerical method to obtain a turn-based high frequency model of transformer windings. This model considers the frequency dependent effects of parameters. In the proposed model, FRA traces are directly derived from a high frequency finite element model (FEM) without employing the complex circuit model. For this purpose, a single-phase transformer is simulated using 3D FEM, which emulate the transformer and FRA measurement operations. First, the model is validated with measurements for healthy state of the windings. Afterwards, various levels of disk space variation (DSV) fault are implemented. This approach will facilitate the precise fault simulation and ease the objective interpretation of FRA.