You Can’t Break the Law of Physics
You can't break the laws of physics. Remembering this can be of help in evaluation of dissolved gas-in-oil data.
In some cases, there have been excessive amounts of hydrogen detected in samples and then suddenly on the next sample, it disappears.
It is possible that hydrogen can escape from the gas blanket of transformers if the transformer leaks or vents periodically. However, given the volume of oil in large power transformers, this cannot occur quickly.
Most often, if there is a sudden change in the hydrogen content for samples taken close together – such as in days or weeks – it is because the hydrogen is really coming from the sample valve. This can be from two sources: one is that dissolved water in oil trapped in the valve can get exposed to low temperatures resulting in condensation; valves can also leak allowing water in.
Either way, free water can break down if there is current flow from dissimilar metals used in the valve compared to the iron in the main tank. The electrolysis of free water results in the formation of hydrogen and oxygen. Another possible source is the use of galvanized valves that can react with the oil to form hydrogen.
Good sampling practice involves wiping out any free water in the main part of the valve and flushing out enough oil through a reducing fitting so that all of the dead oil volume is removed.
The dead volume of oil is that from the tank wall (and sometimes plumbing can extend into electric apparatus), to the sample valve end. The petcock valve should also be flushed. Then the best samples are taken using a reducing fitting on the main part of the valve.
With a good sample that comes from the main tank these potential valve issues are taken out of the equation.
Interested in learning more?
Read about other sources of hydrogen in transformers documented here: Case Studies Involving Insulating Liquids and Materials from the Doble Materials Laboratories