Gradient-based algorithm for the distinction of fault and inrush currents in low power transformers

Abstract

In this work, it is presented an algorithm based on the directional gradient method for distinguishing source surge current, or inrush currents, from transitory fault currents in transformers. Methods and simulation are introduced and the effectiveness of the gradient method is investigated for situations that can lead to a false operational of the protection systems. Three special cases, composed of a combination of external faults with magnetizing inrush current for a $\Delta -\Delta$, 138/13.8 kV - 30 kVA power transformer are studied. The power system transformer is modeled in the software ATP-Draw by the component SATTRAFO. The results indicate that the gradient method is effective for detecting and distinguishing inrush from fault currents when the angle of the gradient of the primary current together with its amplitude is appropriately used as boundary conditions. As a consequence, the algorithm could be used as a fast single/combined one cycle time-domain detection method in order to optimize the 87T protection function.