In general thermal faults are due to overloading and the Trans just gets too hot and detected by some type of a thermal sensor, though the cause could be other than total power, for example current due to poor PF or circulating current (due to system configuration), harmonics, or failure of cooling fans. This protection is to ideally prevent damage to the transformer due to overheating - a few events like this should not be a problem, but repeated overheating will shorten the life. Whereas as electrical is typically looking for actual electrical problems IN the transformer - short circuits, arcing - etc.
You mention you are having dissolved gas analysis done - this is by far the best diagnostic test. However the description "gas concentration is increasing" does not help, you need analysis of the types of gasses being produced and in what ratios. From this you have a good idea of what is going on. http://www.netaworld.org/sites/default/files/public/neta-journals/NWwtr09_Hamrick.pdf - there are certain levels of gas that can be tolerated, but if too contaminated may need to be vacuum degassed or replaced.
At 25 MVA - not too large - but how was it inspected? If it was opened up ( hopefully NOT drained...), you need to make sure the oil has not absorbed too much moisture - to determine this you can have dew point analysis performed and compare the manufacturers guidelines. If caught in time excessive moisture can be resolved - as long as the tank can withstand full vacuum it can be done relatively easily.
Oh - I have seen more then one case of the core not being properly connected - actually connecting to the case in more than one point, this may not cause arcing but can create a lot of circulating current IN the transformer and then overheating - even whit ZERO load. In larger trans the Core-Case bond should be accessible through the hand-hole at the top. This can be tested by lifting the bond and meggering the core to the case.