Client did DGA on 33/3.3kV 4MVA transformer & found potential overheating on the unit. We followed up with thermoscan and found no anomalies. I've attached the images for your expert opinions. Transformer's running on 60% load.

TRANSFORMER_IR.pdf (136 Kb, 105 downloads)

I agree there are no obvious anomalies evident in the images taken at this load level.

Of course as you know this doesn't rule out internal problems at all. There are a wide variety of problems for internal defects that the DGA can pick up... and the possibilities could be narrowed down knowing the specific mix of gases (do you know the gases?). At any rate there are many defects including internal localized hotspots that the DGA can pick up and the external I.R. can be completely completely blind to.

what other condition monitoring method for transformer? best practise?

Partial Discharge in Transformers in Online state using four acoustic sensors on the wall of the transformer and CT on the ground lead.So using both the acoustic and electrical method to confirm of partial discharge in transformer and this happens before we reach the DGA stage.

Is that an IR window at the top portion of the transformer???

What are the DGA results suggesting overheating? Have you re-ran the sample to confirm? Any DGA history?

Also, see IEC 60599, Mineral oil-impregnated electrical equipment in service – Guide to the interpretation of dissolved and free gases analysis

Interesting. Can you post your DGA results? Also, what type of oil - mineral, I would assume, but I have been involved in an increasing amount of work related to silicone oil in transformers, recently.

Also, have you checked the loading on the unit (kVA)? And what types of loads?

Howard

electicpete/howard: will get back on mix of gases/results. It's a mineral oil transformer.
dugan: don't think that's an IR window. DGA has been done on monthly basis I was told.The DGA tester confirmed overheating.
And guys, I'd think when DGA is done, PD should show when detected. In this case, no PD was detected by either DGA or IR and confirmed by PD measurement I carried out (see sensors on cable box).

As E-pete suggests, DGA is typically much more sensitive to changes in oil-filled devices, especially large transformers, than infrared. However, if the internal heating, as confirmed by the DGA, is occuring on a bushing connection, you MAY see heat at the base of the bushing. Clearly in your case you'd need to get up higher to even see the base of the bushing on top of the transformer. Look for a subtle and often diffuse heat pattern. Of course it may also be heating from some other source which with a thermal signature that is difficult or impossible to see.

I would also strongly suggest you reshoot the transformer on an increased frequency and carefully note the loading conditions. Something may show up or you may notice a change. Also, USE A NARROWER SPAN, say 10C; this will cause some of the image to be "saturated" but, by adjusting the level up and down, will allow you to see much more subtle patterns than your current adjustment does.

Yes, I also agree there is nothing too obvious about the images. I agree with John, you may find something with a narrower span, or a different pallette, I would also suggest you try an isotherm with a very small width, and adjust its level up and down, you may find a pattern that is not too obvious to the eye.

You do have to remember there could be substantial heating inside this that may not be possible to see from the exterior, due to a combination of the distance from the source and the oil cooling, so be very careful not to say there is no problem.

Finally as this appears to be located outdoors, I would suggest you re-image it at night, with all the outside lighting off, to completely eliminate the possibility of absorbtion in the visual part of the spectrum, as you could be looking for a VERY subtle pattern.

Hi Nizam, there are diffrent methods used when analysing DGA results i.e Duval triangle methods, absolute value methods, gas ratio methods etc.some of these methods can defect every transformer you have on your system. you infact did well to follow up the DGA analysis with an infrared scan.But when scanning did you consider the weather conditions and actual loadings on the day of the scan and oil sampling day?Its possible your DGA results were as a result of an isolated incident or sampling errors in this case follow up your DGA tests and monitor consistence in the changes in gas levels as apposed to absolute values.Your scan results are not indicating any anormally in the transformer.

Hi guys, managed to get 5 DGA samples done. What can you make out of it?
The TF's outdoor, mineral oil, running on 60% load & it was pretty cloudy when the scan was done.
Could air lock on the TF fins be a factor of internal overheating?

TF_DGA1.pdf (132 Kb, 36 downloads)

Based on the IEC 60599 method, your data suggests a thermal fault exceeding 700C (thermal fault in oil) - a severe fault.

Do you have any similar transformers that you can compare IR scans with? Compare temperatures / loads, etc.?

A quick look using the key-gas method tends to confirm what Jamie said.

Also since CO and CO2 are not going through the roof (even though they're fairly high), one guess is a hot connection at a non-insulated location.

One possible thing to try would be taking the transformer off-line for a few quick easy checks:

Doble test including winding resistance and TTR. Note in the process of performing TTR on all taps, you may wipe free any a high-resistance contact at the NLTC. Core ground test if accessible.

One thing to note is that the gases appear fairly steady. If you are doing detailed analysis on the DGA, you would try to figure out if this represents an equilibrium level with continued input of gases (an on-going fault), or a one-time event that put created the gases and the residual gases are remaining with no further gas generation (no remaining fault). Of course if you can trace the pattern further back in history it may be more obvious whether the gases suddenly appeared in a one-time event or slowly built up to equilibrium. That type of analysis would need to consider the type of preservation system (conservator, N2 blanket, N2 regulator etc). Also if the transformer has previously been drained/processed, it is helpful to study the pattern followiong any draining/refilling in the past. From the relatively high variation of nitrogen and oxygen, it is certainly plausible this transformer does a lot of breathing and you may be seeing an equilibrium level with continued generation of gases (an ongoing fault).

Another small clue that this transformer is losing gas from the oil is the fact that H2 is listed as nondetectable (ND).

We don't know what the threshhold of detection is, but since some gases show at 1ppm, I assume 1 ppm is min detectable and H2 apparently has even less than 1 ppm (!).

This is extraordinarily low, considering the other gases. H2 is produced in all the scenario's that produce the other hydrocarbon gases.

H2 is also the first to leave the oil into the gas space since it has the lowest Ostwald solubility coefficient.

The only possible conclusion is that H2 has left the oil into the gas space and possibly into the atmosphere. So there is some very noticeable loss of gases from the oil which has occurred or continues to occur within the this transformer due to gas conditions above the oil.

Again supports the conclusion that there might be ongoing gas introduction from a fault which is being matched by gas loss into the gas space and possibly atmosphere, giving a relatively constant equilibrium level in the oil.