Back to Main
The Case of the Gassing Transformer
by Dave Huffman
Power Systems Testing
This bulletin addresses a problem that my company found to occur frequently in a particular piece of equipment. After many years, we found the answer to a question we often posed to the industry.
It is likely there is a group of users who are currently dealing with this
problem and who are unaware of its solution. I hope the solution described
in this article will bring an end to what could otherwise be an expensive
For more than seven years, we tracked oil samples of seven substation class liquid-filled power transformers and observed the gas in the oil increase with time. These units are 2500 KVA 12,470 volt delta primary. Three of the units are 4160 volt wye secondary; the other four are 480 volt wye. For the first few years, the increase in gas was moderate. Some correlation was made to a steady increase in load. But then the rate of increase began to climb.
When the gasses reached over ten percent total combustible gasses (TCG) the oil was replaced. Within a year the gasses were at all-time highs. But the gas combinations never completely fit any particular Roger's ratio code, and the electrical tests revealed nothing out of the ordinary. Megger and power factor results over time really didn't change much, certainly not to the degree that the oil results did. The predominant gas was hydrogen which had reached levels that caused the lab technicians to call wanting to make sure it was a sample from a transformer and then wanting to advise me of the very high levels (which was appreciated).
We contacted the manufacturer as well as Doble Engineering for evaluation of our results. We also wanted to determine whether or not our problem was unique. The manufacturer reported no known problems like this; Doble had several suggestions all of which were pursued except for one, which was checking the core ground. Checking the core ground was not possible, since the transformer was internal.
Degassing was the next step taken, with positive results at first. But within a month all units were well on their way to previous levels. Although the secondary busses for each voltage could be tied together, the 480 volt loading was such that none of the 480 volt units could be removed from service; one of them desperately needed to be removed. A spare unit was purchased to replace the failing unit during a shutdown.
Now knowing the cause of the problem, the customer developed a system to tighten the bolts without removing the transformer. Using a boroscope and a very long ratchet and socket, two people could navigate the tools to the bolt by way of the inspection plate on top of the unit. All they needed was a several-hour outage and more than a little patience. After each unit was repaired and cycled temperature-wise, the gasses came out of solution and were allowed to escape to atmosphere. Recent DGA samples show gas levels returning to normal. We thought we'd pass this along to the industry just in case anyone is working on a similar problem.