Inrush on a reverse fed step down transformer

magneticstransformer

I found a paper from ABB with some unexpected information on inrush current. First, they feed 480V to the high voltage side of a 480V/208V step down transformer and measure the inrush. Then, they feed 208V to the low voltage side of the same 480V/208V transformer – using it as a step up transformer. Measured inrush current in the second case is much higher than the ratios of the transformer would predict. Can any transformer experts explain the asymmetry in this experiment? Maybe the leakage inductance is substantially different between the windings and plays a big role in inrush?

The primary and secondary full load amps of General Electric 9T23B3874 step-down transformer are 90 amps @480 VAC and 208 amps @ 208 VAC. When connected step-down and energized at 480 VAC, the maximum peak inrush current is approximately 990 amps or 11 times the rated 90 amp primary winding full load current. But when connected step-up and energized at 208 VAC, the maximum peak inrush can reach 7700 amps or 37 times the rated 208 amp secondary winding full load current.

Best Answer

The inrush current (considering in p.u. of rated current) will be different for each winding in which the transformer is energized. The question is not really related to "if the transformer is step-down or step-up," but rather, with the diameter of each winding. Of course, usually the low voltage windings are wound closer to the core, then the inrush current is higher for these inner windings.

This is because during the inrush event the core is driven to saturation and the current is limited by the energized winding "air-core reactance". If you look for the reactance formula of a coil, you will see that the larger the area, the higher the inductance/reactance, therefore, more "internal" is the winding, lower will be the air-core reactance and larger will be the inrush current.

It's not exactly related to this question (which winding of a transformer has the bigger inrush current), but in this post I've described a model for inrush current calculations in LTSpice.