How is transient voltage related?
While no single factor would positively assure longevity for a transformer, the combination of several things seems to be very effective.
We left off last week discussing some of the possible reasons for differences in reliability of liquid transformers versus dry-types, when primary windings are switched by vacuum circuit breakers (all my personal opinions, again). Here is probably the most important reason of all:
In a liquid transformer, it’s not difficult to add Basic Impulse Level (BIL) directly into the primary windings. All of the more than 500 liquid transformers I’ve commissioned were specified and ordered with primary windings having a BIL at least one level above ANSI standard, and often TWO levels above ANSI and NEMA. Any unit having a primary voltage class of 25 kV, for example, would have a primary winding BIL of 150 kV or 200 kV (versus 125 kV ANSI standard). That extra 25 kV or 75 kV margin of BIL can make all of the difference in enabling a transformer to survive and get along for a full, long life in that tough world of data center transformers out there.
Many of the IEEE papers I’ve read and studied about transient voltage failures suggested that adding BIL to the windings won’t fix the problem. I think that’s true, but only within relatively narrow ranges. If someone could design and build a medium-voltage (MV) distribution transformer that had a Basic Impulse Level of, say, 1 MV, I’m pretty sure this entire problem would go away.
The vast majority of dry-type transformer failures in data centers have occurred on transformers having primary windings of 15 kV class. I thought for many years that the reason for this was that the vast majority of distribution transformers installed in data centers happened to have primary windings of 12.47 kV or 13.2 kV or 13.8 kV. In other words, the population of 15 kV class units installed was far greater than other voltages, so simple statistical probability would suggest that 15 kV primary units would fail most frequently.
I’m no longer so sure about that. The majority of early failures I investigated, way back in the 1980s, were dry-type units having 15 kV primaries, with a primary BIL of only 60 kV (which, incredulously, STILL remains the IEEE standard BIL for dry-types to this day, as shown in IEEE C57.12.01-2005 Table 5). And, the majority of those were switched by vacuum breakers that had severe current-chop characteristics, before the relationship of the metallurgy of the main contacts in vacuum breakers was really understood.
The electrical industry soon wised up. Vacuum breaker contact metallurgy was quickly improved, and consulting engineers began specifying at least 95 kV BIL or 110 kV BIL on 15 kV-class primary windings. With these changes, the rate of failures diminished quickly—but still remains too high.
Again, I have no hard casualty statistics, but the majority of failures that have occurred in the last 10 years have been units with 15 kV-class, 95 kV BIL class windings. I’m not aware of any failures of any units having 150 kV or 200 kV BIL primary windings, and only a few failures on units with 125 kV BIL windings.
I now think that this has to do with the fact that the transient voltage that appears across the transformer winding during breaker operation is completely unrelated to the actual nominal voltage of the system to which the transformer is connected. The transient voltage is instead a function of the magnitude of the chopped current, and the effective capacitance of the winding itself.
Dr. Allan Greenwood, one of the pioneers in the development of vacuum circuit breakers, showed that it is easily possible to achieve a single shot of transient voltage on breaker opening exceeding 130 kV across a typical 13.8 kV winding, with just 2.5 amps of current chop. It makes no difference whether the system voltage is 5 kV or 15 kV or 25 kV or 35 kV class—if the winding capacitance and the magnitude of current chop are the same, then the transient voltage across the winding will be the same. This suggests that present standards for 95 kV or 110 kV BIL in 15 kV units could be marginal, and that the 60 kV standard for dry-types seems woefully inadequate.
While no single factor discussed above would positively assure longevity for a transformer, the combination of all these things (lengthening the primary cables, connecting MOVs directly to the windings, increasing the winding BIL, and immersion of windings in fluid inside a sealed tank) seems to be very effective.
There is no exact science here, and I’m not suggesting that a liquid transformer installed in this manner can NEVER experience a winding failure due to transient voltages from upstream switching operations. I’m only suggesting that, in my experience, the probability of a primary winding failure can be reduced to very close to zero with a properly designed liquid transformer installation within a good system.