asi DITION Messer Icy Advancements in Dissolved Gas Analysis: Accounting for Gas Loss Dr. Zachary H Draper & Dr. James J Dukarm Uarsomes that be aun pees themsees do nt ase fase bute st by pres are hat mg fate ansfrmer Ihe arse fses tags measured gas production is less than diagnosis may be inaccurate, Conventional DGA assessments usually rely on a limits based approach where higher concentrations of dissolved fault gases are assumed to represent a worse condition. Thete are numerous reasons why fault gases in e transform= er might be lost to the atmosphere, either on purpose or by accident. For example, some transformers are free-breathing by design. Othe ‘ot supposed to be free-breathing may exchange gases with the atmosphere via a ruptured con- servator bladder, a leaky bushing gasket, orloose fittings. N trogen blanketed transformers shed headspace gas to relieve high pressure and then later make up the lost gas with pure nitrogen, Because of the lowered fault gas concentrations in the headspace, some dissolved fault gases are released from the oil to re-establish equilisrium with the heedspace. This results in a decline in dissolved fault gases as the transformer goes through thermal cycling and headspace pressure reg lation. The loss of fault gas can be much w former is leaking headspace gas, Be sure to track nitrogen consumption to identify leaking transformers One way to compensate partially for gas loss is to calculate the overall increase in each fault gas concentration by adding Lp positive increments between oil samples over the trans. former's history, ignoring negative increments. Itis usually not possible to knaw how much faul the use of cumulative gas concentrations is helpful to detect, problem and provide a low ball estimate ofits severity. The figure illustrates a nitrogen blanketed transformer that had three gassing events, but due to gas loss between, se ifthe trans- 15 was lost, but often {HV ELECTRICAL TESTING SPECIAL EDITION ‘the events, the abserved gas concentrations (blue) stayed! below IEEE limits. Since the gas levels never exceeded any limits, the utility was surprised when tutn-to-tuen arcing tripped the transformer. A post mortem inspection revealed extensive charring of the paper winding insulation is evident from the cumulative gas trend (redi that the transformer had a thermal problem affecting paper spanning several years, When using cumulative gas concentrations for DGA, Interpretation, itis very often necessary to smoot t data to reduce the amount of sample-to-sample noise due +0 sampling and measurement uncertainty. Otherwise, the upward component of noise will contribute to the cumulative data, raw Watch for more articles in this Electricity Taday series about advancements in DGA interpretation. ‘Accumulating Fault Gas Generation Dr Zachary Draper a scientist with Delta-X Research Ine, His current reseorch focuses on reliability statistics, mult-dimen sional gas diagnostics for transformers and load tap changer and seasonal variability and signal processing for online DGA imonitars. He also provides training an how ta interpret ad ry-based dissolved gas analysis. IGRE, Dr Draper eceived @ BSc.an ysics ond Astronomy from the University of Washington, and Ph.D. in Astrophysics rom the University of Victoria, Far more vanced analytics and reliabil member ofboth i Information on this article, please email infowdeltoxreses orvisit vnvwsettaxresearch com. FREE Subscription: wunweleetricty-today.com