Non-invasive generator diagnosis: A model-based idea evaluation

Subhashish Dasgupta1, Simi Valsan2, Ranjeet Kumar2, Arinjai Gupta2
1ABB
2ABB
Published in 2023

It is an ever desirable and challenging engineering endeavour to develop non-invasive systems for rotary machine diagnosis. Contrary to invasive methods, a non-invasive diagnostic method offers the advantages of avoiding stoppage of the machine for repairing the diagnostic system and interference with the functionality of the machine. Also, very often the space required for instrumenting sensors within the machine is limited. In this research work, a multiphysics model of the hydro-generator was developed using the COMSOL multiphysics code. The model solved the maxwell’s equations across the stator and the rotor domains, using information like the rotor current, rotational speed of the rotor and the external capacitive-inductive load circuit. The model computed the magnetic flux within the generator domain as well as the leakage flux in the surrounding air external to the domain. Besides flux, the induced voltage and current in the stator windings were also calculated. The purpose behind modeling of the leakage flux outside the domain was to obtain a proof-of-concept behind the idea of detecting defects within the generator by using the transient profiles of the leakage flux measured at multiple points. While non-invasive sensing in rotary machines is a topic investigated by past researchers, the idea of employing multiple external flux sensors to not only detect defects within the machine but also to identify specific defects like stator and rotor imbalance or short circuiting is relatively new. Analysis of the flux profiles, revealed that each defect had a corresponding unique flux vs time profile, characterizing the defect. To our knowledge such an extensive set of findings have not been reported in literature.