Responses to some questions, more on snubbers
Over the course of this series of discussions, I have received a number of questions and comments from engineering firms by email.
Over the course of this series of discussions, I have received a number of questions and comments from engineering firms by email. Three or four of these asked similar questions, and the response to that set of questions helps to segue nicely into the discussion to follow below.
The questions went something like this: “You wrote earlier that you believe it makes sense to specify the highest BIL you can get in the primary winding of a medium voltage distribution transformer, regardless of the type of construction of the transformer. Please tell me why, as a specifying engineer, I would want to specify a higher BIL for the transformer winding than for the medium-voltage metal clad switcher that supplies the transformer?”
An entirely reasonable question. Let’s try it this way. It’s easy to think of the entire problem of failures of dry-type transformers from switching-induced transients as being due to a voltage surge that hits the transformer winding from the outside world. Most people tend to think of the problem exactly this way, and would attempt to add transient voltage surge suppression, just like you would do in order chop a utility switching surge or a lightning strike coming into the transformer from an upstream utility system.
But, that is not what’s happening here at all. The vast majority of catastrophic dry-type transformer failures I’ve seen were pretty simple events, and many of them occurred during commissioning of the data center, often during a “pull the plug” test.
The plant is running in a normal steady state condition, with some magnitude of load on the transformer, and the upstream primary breaker is opened to simulate a “utility failure.”
When the breaker is opened, the primary winding becomes almost instantly disconnected and isolated from the switchgear. The switchgear is now out of the picture - it did its job, and the IEEE standard 95 kV BIL that you specified for the switchgear was just fine.
But, even though there is no longer a voltage source connected to the transformer from the outside world, the primary winding of the transformer has trapped a lot of energy inside, and since the current flow through the winding was stopped nearly instantly, that trapped energy escapes by shooting out the ends of the windings as an extremely high voltage surge across the outer winding terminals.
One of the main purposes of R-C snubbers is to fake the transformer primary winding into believing that it contains more capacitance than it really does. This transient voltage originates within the winding itself, even without a source of external voltage connected to it, and manifests itself as a huge voltage across its outer terminals. If that voltage has a magnitude of, say, 120 kV, and the winding has an impulse rating of only 95 kV, the winding can easily destroy itself. The rate of rise of the voltage is nearly a vertical line, from nothing to peak value within microseconds – a dV/dt approaching infinity is some cases.
With the magnitude of the voltage defined by V = i x √(L/C), it’s apparent that the lower the capacitance, the larger the voltage, and that the greater the capacitance, the lower the voltage. (This also illustrates why it’s desirable to use a vacuum breaker that has low current-chop characteristics, because reducing current chop by half will reduce the magnitude of the voltage transient by half).
This is a simplified representation that covers only one type and one sequence of switching induced transients, but it’s the most common type – and it illustrates why you would want a much higher BIL rating in the winding that in the upstream switchgear.
In most cases the failure of the winding in this condition goes temporarily unnoticed. It’s usually a fairly calm and quiet event. The very loud and bright catastrophic failure generally doesn’t occur until the upstream breaker is reclosed, and the all of the available energy from the source can flow into the faulted winding - usually producing a spectacular (and often frightening) result.
More on snubbers
Last week we discussed some concerns about mounting R-C snubber components within substation transformer enclosures. Personally, I see several drawbacks in mounting so many components having uninsulated live parts operating at 15 kV, 25 kV, or 35 kV in such close proximity to each other, and in close proximity to the core-and-coil of a dry-type transformer. To summarize these again, many such installations leave inadequate clearances between live parts on opposite phases, and between phase-to-ground, such that flashovers can occur all too easily. Also, the heat from the core-and-coil can cause excessive ambient air temperatures surrounding the capacitors and Metal Oxide arresters, causing them to age prematurely. (One rule of thumb is that the life of a typical capacitor is cut by half for every 10 C increase in case temperature above about 60 C, and most MOV’s are limited in their ratings to an ambient temperature of 60 C max).
David Shipp, P.E., of Eaton Corp Tech Center, sent me some photos of an R-C snubber arrangement Eaton came up with, that I believe solves all of these problems.
The very obvious benefits of this R-C snubber assembly are:
- It’s contained entirely within the plan outline of the transformer, so that it adds no footprint and requires no additional floor space in an electrical room.
- Being self-enclosed, it allows two full steel barriers between snubber components and the transformer core-and-coil. A snubber failure can’t damage the transformer, and a transformer failure can’t damage the snubber.
- It allows a ventilating air space between the transformer roof and the snubber enclosure to promote cooling of the snubber components, that will extend life of those components.
- Since it’s located above the transformer, it allows use of very short tap leads to connect the snubber to the transformer line terminals.
I have seen installations where a large, separately-enclosed. free-standing snubber assembly was mounted against a wall across the room from the transformer, and interconnected by shielded MV cable having a total length as great as 25 ft. That arrangement diminishes the effectiveness of the snubber - remember, the objective here is to add capacitance directly to the winding itself. (The same principles as why you always specify TVSS in low voltage switchgear to be connected directly to the bus with large and very short cables).
The overall arrangement appears to me to be the ideal method for applying snubbers on dry-type transformers. I’d expect that manufacturers of transformers will love the arrangement as well, as it relieves them of the responsibility of providing snubber components and figuring out how to mount them inside their enclosures, and lets them focus on their core competencies.
I can’t resist offering this advice: If you’ve decided to specify dry-type substation transformers in your data center, and you have reached the big decision to “snub”, then PLEASE snub safely and responsibly.