Core Loss Prediction in Hydroelectric Generators and Analysis of a Contra-Rotating Machine Topology

Sammanfattning: The call for increased accuracy in the process of design and analysis of electrical machines has encouraged the use of numerical methods to calculate the electric and magnetic field distributions within this type of apparatus. The finite element method is a numerical technique that has become specifically widespread and popular, due to the relative swiftness and accuracy with which it is possible to treat complex machine geometries and non-linear material characteristics. From the calculated spatial and temporal distributions of the magnetic field inside the machine, the induced electromotive force, the cogging torque and the air-gap torque at rated load operation can be derived. Moreover, the power losses in the iron parts of the machine can be estimated if loss data for the lamination materials used in the poles and in the stator core of the machine are available. This thesis comprises two separate studies which fundamentally rely on numerical field formulations of the problems at hand. In the first study, the characteristics of a radial flux permanent magnet generator with two concentric contra-rotating rotors are explored. It is found that common core paths will give rise to a pulsating inter-rotor flux disturbance, which upsets the three-phase symmetry in such devices. The disturbance will however not be of any practical importance for synchronous generator operation, provided that there is a screening iron core between the rotors. The second study addresses the problem of core loss prediction in large hydrogenerators. Here, a special emphasis is on the importance of the additional rotational loss, attributable to bidirectional fields. It is found that model corrections introduced to account for dynamic and rotational effects typically increase the core loss prediction with about 28%. Furthermore, the magnitude of the additional rotational loss was found to be correlated to the stator slot dimensions.