Gas Limits in transformer
Table 21 shows IEEE limits , compared with Doble  in a study of 299 operating transformers. The table of gases from the Doble study seems more realistic, showing gas level average of 95% of transformers in the study. Note, with the last four gases, limits given by the IEEE (trial use guide) run over 70% higher than the Doble 95% norms. But with the first three gases, hydrogen, methane, and ethane, the IEEE limits are well below the amount of gas found in 95% norms in the Doble study. We obviously cannot have limits that are below the amount of gas found in normal operating transformers. Therefore, it is suggested that we use the Doble (95% norm) limits. The 95% norm limit means that 95% of the silicone oil transformers studied had gas levels below these limits. Obviously, 5% had gases higher than these limits. These are problem transformers that we should pay more attention to.
In table 22, the IEEE limits for L1 were chosen. For L2 limits, a statistical analysis was applied, and two standard deviations were added to L1 to obtain L2. For L3 limits, the L1 limits were doubled. Limits L1, L2, and L3 represent the concentration in individual gases in ppm. G1 and G2 represent generation rates of individual gases in ppm per month. To obtain G1 and G2 in ppm per day, divide the per month numbers by 30. Except for acetylene, G1 is 10% of L1, and G2 is 50% of L1. The generation rates (G1, G2) are points where our level of concern should increase, especially when considered with the L1, L2, and L3 limits. At G2 generation rate, we should be extremely concerned, reduce the DGA sampling interval accordingly, and perhaps plan an outage, etc.
Except for acetylene, generation rate levels G1 and G2 were taken from IEC 60599  which is used with mineral oil transformers. Any amount of ongoing acetylene generation means active arcing inside the transformer. In this case, the transformer should be removed from service. These criteria were chosen because of an absence of any other criteria. As DGA criteria for silicone oils becomes better known and the quantified table 22 will change to reflect new information.
As with mineral oil-filled transformers, gas generation rates are much more important than the amount of gas present. Total accumulated gas depends greatly on age (an older transformer has more gas). If the rate of generation of any combustible gas shows a sudden increase in the DGA, immediately take another oil sample to confirm the gas generation rate increase. If the second DGA confirms a generation rate increase, get some outside advice. Be careful; gas generation rates increase somewhat with temperature variations caused by increased loading and summer ambient temperatures. However, higher operating temperatures are also the most likely conditions for a fault to occur. The real question is, has the increased gas generation rate been caused by a fault or increased temperature from greater loading or higher ambient temperature?
If gas generation rates are fairly constant (no big increases and less than G1 limits above), what actions need to be taken if a transformer exceeds the L1 limits? Begin to pay more attention to that transformer, just as with a mineral oil transformer. It may be necessary to shorten the DGA sampling interval, reduce loading, check transformer cooling, get some outside advice, etc. As with mineral oil transformers, age exerts a big influence in accumulated gas. Be much more concerned if a 3-year old transformer has exceeded the L1 limits than if a 30-year old transformer exceeds the limits. However, if G1 generation rates are exceeded in either an old or new transformer, the level of concern should be stepped up.
If accumulated gas exceeds the L2 limit, it may be wise to plan to have the transformer de-gassed. Examine the physical tests in the DGAs and compare them to the Doble/IEEE table (table 23) (Reference Book on Insulating Liquids and Gasses) . The oil should be treated in whatever manner is appropriate, if these limits are exceeded.
If both L1 limits and G1 limits are exceeded, be more concerned. Reduce sampling intervals, get outside advice, reduce loading, check transformer cooling and oil levels, etc. If G2 generation limits are exceeded, be extremely concerned. It will not be long before L3 limits are exceeded, and consider removing the transformer from service for testing, repair, or replacement. If acetylene is being generated, the transformer should be taken out of service. However, as with mineral oil transformers, a one-time nearby lightning strike or through fault can cause a “one-time” generation of acetylene. If you notice acetylene in the DGA, immediately take another sample. If the amount of acetylene is increasing, an active electrical arc is present within the transformer. It should be taken out of service.
With critical silicone (or mineral oil-filled transformer), such as a single station service transformer or excitation transformer, a spare should be available at another facility. If there are no other possible spares, consider beginning the budget process for purchasing a spare transformer.