[Back]

G. Shilyashki, H. Pfützner, P. Hamberger, M. Aigner, E. Gerstbauer, G. Trenner:
"Numerical MACC-Modeling of Local Peak-to-Peak Magnetostriction Distribution in a 3-Phase Transformer Core Package";
Talk: International Conference on Engineering Vibration, Ljubljana, Slovenia; 2015-09-07 - 2015-09-10; in: "ICoEV Book of Abstracts", (2015), ISBN: 978-961-6536-96-7; 106.

For the estimation of regional distributions of magnetostriction (MS) in transformer cores, the current paper proposes a novel methodology. It is based on a four-step procedure that is summarized in the following: (i) Local induction values of the transformer core are calculated by a novel Multidirectional Nonlinear Equivalence Circuit Calculation (MACC) method. The method considers the multi-directional non-linearity and anisotropy of the material. Local induction values in a large number of flux paths can be determined no only in rolling direction (RD), but also in transvers direction (TD) in order to simulate rotational magnetization and also in diagonal direction (DD) in order to consider the effects of the overlaps. (ii) A look-up table is established for MS-values for the given type of material for different magnetization conditions. The values are obtained from catalogue data or measured e.g. by means of Single Sheet Tester or Rotational Single Sheet Tester. (iii) The corresponding MS-values are assigned to each investigated flux path, taking into account the peakinduction in RD and TD. This yields a 2D map of magnetostriction (3D maps being possible as well). (iv) Numerical Integration is performed for individual directions, e.g. for the axis of the yoke, in order to estimate the global displacements of the core. The method was tested for a 3-phase transformer core package stacked from GO material C130-30 with outer dimensions of 1200 mm. Minimum values below 1 ppm resulted for the limbs. As to be expected maximal values up to the order of 10 ppm resulted in the T-Joint due to rotational magnetizaton. The results are in rough agreement with experimental data of model cores.