Physical Surrogate for Large Parametrized Loosely Coupled Multiphysics Problems: Application to Power Transformers
Please login to view abstract download link
Power transformers are a critical asset in power distribution systems. Their accurate physical representation is key in the construction of a digital twin enabling their health monitoring. The lifespan of a power transformer in the long term is mainly associated to its hotspot temperature. Its location and temperature are dependent on the design of the transformer’s electromagnetic circuit. Our aim is to be able to represent a large array of such circuit designs. Therefore, we need to alleviate the computational burden associated with large scale, multiphysics parametrized models. We propose a physics domains separation approach. We leverage the loose coupling between the electromagnetic system and its cooling system to separately represent and reduce the computational cost associated with the parametrized nature of the active part of the transformer. We further develop a reduced order model of the electromagnetic circuit for efficient simulation in parametric studies.