Post DGA of Generator Transformer after Fire

Your first and best source of DGA (Dissolved Gas Analysis) is the firm that has been analyzing your samples. Without the regular interval sampling to form a baseline, it will be a challenge to attribute a cause to the DGA because the fault and subsequent fire may have distorted the post-fault sample. If you have a series of pre-fault analyses then you can look back and see which parameter led to the fault.

Good answer, RAM.

Oil analysis, and DGA in particular, does not have bright line "traffic signal" limits. The poster needs to know the condition of the oil before the fault (and how the parameters were trending) in order to determine if the fault caused additional internal problems.

You do not say in which city or country this happened. You do not say how much historical data exists. You reference IEEE, which is a good thing. There is a lot of data out there on electrical equipment forensics. There are folks you can hire to help. Experience counts.

There are some electrical equipment failure types that can be successfully analyzed forensically on data gathered after the fact. Most cannot be. I have dissected very large power transformers, to study the way that heat damaged their internal components. This was valuable because we had many years of data on the changes in Furan production, which resulted from heat damage to their cellulose. We knew the units were being heat-damaged, but operating them at 'overload' was profitable. By carefully taking them apart, we were able to pinpoint the areas of greatest heat damage, and how this could have been prevented by better design and manufacturing.

There is great value to the regular analysis of the oil in a transformer and in bushings. Trends can predict most failures. Smaller units may not get this analysis done. Small units and non-oil-filled equipment tend to have 'family experience' that can indicate future failures. If you know that a certain brand and date of manufacture of lightning arrestor has a rainwater leakage problem, you can test and replace all you own. The oil in a bushing will show some types of problem, but its electrical properties will degrade over time. When you see losses climb, you can schedule a bushing replacement. Some big bushings have oil sample ports. Some have ports that allow air to leak in below the electrically stressed zone. If you sample there, and air bubbles in, the unit can blow up the instant it is re-energized, killing folks nearby. Do not allow air to get into such a zone without properly vacuum treating the whole bushing. Open a port above the oil level, and use a small tube to get your oil sample, in such cases. In some cases, there is a sealed 'weather cap' over the actual oil sample port. In such a case, you can get or make a tool to determine pressure outside the oil port cap, to determine if there is a partial vacuum, indicating that air might leak in dangerously. I made one with a bit ingenuity, brass and o-rings.

Before the fire, the concentration of fault gases in the oil, and of furans in the oil, if it had hot paper, can foretell problems. After the fire, there will be so much new fire debris and byproduct everywhere, that analysis of the oil will probably tell you nothing other than that you had a fire.

Did the bushing failure lead this series of events? Look at the brand and age of the failed bushings. Do you have more of the same brand, model and age in your system? If you have annual oil analysis on all your bushings, look for patterns. Climbing concentrations predict failure. IEEE publishes gas ratio trends and limits.

Replace any bushings which have bad data, or from 'troubled families'. Bushings can be rebuilt in a shop. An oil change may be all you need. Maybe bad gaskets are letting water in. Some designs allow for gasket change. If there aare high losses that do not go away with an oil change, odds are they should be scrapped, unless of an easily dissassembled design. If bad stuff has built up inside the paper, the bushing should be replaced.