The short answer is no, primary windings of liquid-filled transformers don't fail. Here's why.

Joe Guentert; Owner, Power Distribution Systems

Share

In the same 35-year period of time, I’m counting up the number of failures of primary windings of liquid-filled transformers I’ve experienced, or have been reported to me, or that I’ve heard about or read about, caused by vacuum breaker switching in data centers. It’s exactly NONE.

In that same period of time, I’ve commissioned well over 500 liquid-filled transformers installed indoors and outdoors at large data centers, and have experienced zero primary winding failures, despite hundreds of deliberate switching operations of upstream vacuum breakers during commissioning and later plant operations, under all possible conditions of system loading and connection configurations in various primary loop arrangements. Nearly half of these had 34.5 kV primaries, as nasty and tough and ugly a utility distribution voltage as there is on the planet (more to come about system voltages and their relationship to the problem).

Again, I’m only relating my personal experiences and observations and opinions here. If anyone knows of failures of primary windings in liquid transformers installed in data centers caused by this phenomenon, I would like very much to learn about them.

Some Explanations for the Differences 

Clients ask me, “if you believe all of this so strongly, what logical reasons are there to explain these differences?”  Here are some of the reasons:

• Of the 30 or more failures of dry-types I’ve personally investigated, approximately 20% of those were failures deep within the windings, suggesting oscillatory transients at the resonant frequency of the transformer, most likely caused by re-strike ignition phenomena during breaker opening at light transformer loading. (All of those failures coincidentally had secondary loads of 6 or 12 pulse input rectifiers in the front ends of static UPS systems).
• But, the large majority of the failures I investigated occurred at the ENDS of the primary windings, or at delta corner jumper leads, or at tap connections, and often flashed over to the grounded steel enclosure or to grounded core steel. I’ve come to believe that a major part of the problem was the connections to these live parts were insulated only by air, and that arrangement provided an all-too-easy flashover path on seeing a sudden big blast of L(di/dt) come shooting out the ends of the windings.
• Moreover, these failures occurred in areas where winding insulation was changing from paper to air-only, and where the line impedance was also chancing. Had all these connections been instead immersed under a high dielectric strength insulating fluid, I believe that many of these failures would not have occurred.
• In other words, I think that a number of these failures might not have actually been WINDING failures. They were more likely terminal connection failures due to the inadequate dielectric strength of the surrounding air, and the actual winding damage that did occur might have been mostly just collateral damage. (Interestingly, the majority of these did not even have surge arresters of any type connected to the winding terminals). More discussion about this coming next week.