Vibration Model for the Detection of Mechanical Faults Within Windings and Magnetic Core of Power Transformers
2019
Jānis Mārks

Defending
19.12.2019. 11:00, Rīgas Tehniskās universitātes Enerģētikas un elektrotehnikas fakultātē, Āzenes ielā 12 k-1, 212. auditorijā

Supervisor
Sandra Vītoliņa

Reviewers
Jānis Dainis Dirba, SVETLANA ORLOVA, Nick Papanikolaou

A vibration model is developed in the doctoral thesis that is based on the mathematical modelling of vibrations caused by the transformer due to electrodynamic forces existing in the windings and magnetostrictive effects occurring in the magnetic core. A mechanical configuration of the active part of a large transformer is created within this model, which creates equivalent vibration values for the obtained results on the surface of the transformer tank. Therefore, it is possible to simulate and localize the existing mechanical defects within the active part of the transformer by using the vibration data on the tank surface. Magnetic field model of the corresponding transformers is developed and implemented in COMSOL software for successful operation of the transformer vibration model. It provides information about the values of magnetic induction of the transformer magnetic field in the windings and the magnetic core as well as the current density values in the transformer windings. A system of masses and springs is created within the developed vibration model, which is required to simulate the mechanical movements of the transformer winding and the magnetic core under the influence of electrodynamic forces and the magnetostrictive effect. This system is created for it to be possible to simulate vibrations in both the two-dimensional space needed to simulate vibrations caused by windings and the three-dimensional space designed to simulate vibrations generated by the magnetic core. A dynamic genetic algorithm is applied to find the equivalent vibration values of transformer windings and magnetic core simulation. The stiffness coefficients of the springs of mass and spring system are altered to find the configuration of this system as a result of this algorithm, which produces equivalent values to those recorded on the surface of the transformer tank. The operation of this algorithm is based on introduction of random changes to the system configuration. The positive effects of this process are supported, but changes with negative effects are deleted. A dynamic genetic algorithm is needed because the configuration of the created mass and spring system does not have any correlation with the simulated vibration values and the number of possible variations is too large to simulate them all. The developed vibration model is used to obtain results for 5 large power transformers. These results include conclusions for individual regions of their active parts and visualization of vibration approximation. These transformers are selected with different vibration values on tank surface, which in many cases exceed or are close to the specified limits. This approach allows to verify the correct operation of the transformer vibration model in different cases.


Keywords
Transformatori, bojājumu diagnostika, skaitliskā simulācija

Mārks, Jānis. Vibration Model for the Detection of Mechanical Faults Within Windings and Magnetic Core of Power Transformers. PhD Thesis. Rīga: [RTU], 2019. 117 p.

Publication language
Latvian (lv)
The Scientific Library of the Riga Technical University.
E-mail: uzzinas@rtu.lv; Phone: +371 28399196