TRANSFORMERS AND SUBSTATIONS

Using Gas-in-Oil Standards to Improve the Accuracy of DGA Results and Diagnosis

Dissolved gas analysis (DGA) is widely used to monitor the condition of transformers and other pieces of oil-filled equipment in service. Wrong diagnoses may be drawn, however, if the DGA results coming from the lab are not accurate. For instance, an electrical fault may be mistaken as a thermal fault, which may have serious consequences for the equipment. Several CIGRE and IEC round robin tests (RRTs) have shown that the repeatability (precision) of DGA labs is generally good, but that their accuracy (deviation from true value) is often poor.

The only way to reliably determine accuracy is to run gas-in-oil standards. The procedure for preparing gas-in-oil standards consists in dissolving known amounts of gases in degassed oil [1]. This procedure has been incorporated in IEC and ASTM Standards [2,3]. The most recent RRTs organized by CIGRE using gas-in-oil standards are summarized in [4] and in a more complete form on the CIGRE website (www.cigre.org). The results obtained from these RRTs indicate an average accuracy of ± 15 per cent at medium gas concentration levels (10-100 ppm), for the 25 international participating laboratories. The corresponding (intra-laboratory) repeatability of participating labs was ± 7 per cent on average at these gas levels. A typical example of CIGRE-RRT results using DGA gas-in-oil standards is indicated in Table 1. Only gas extraction methods used in North America are indicated in this Table. Head space and Partial Degassing methods correspond to Methods C and A of ASTM D3612-01, respectively. The Shake Test method is described in [5].

Some laboratories are, of course, more accurate than average and others worse, as is well known in the industry. Laboratories which did not participate in the RRTs can determine their own accuracy, and eventually work to improve it, by running gas-in-oil standards. For most routine labs, however, preparing gas-in-oil standards is time-consuming and costly, and requires highly skilled personnel dedicated to this preparation.

Fortunately, gas-in-oil standards for DGA are now available commercially from Morgan Schaffer Inc. in Montreal, Canada [6]. These gas-in-oil standards are prepared following the specifications and basic concepts of ASTM and IEC Standards and contain certified amounts of dissolved gases. They have been recognized and used successfully during several CIGRE and IEC-RRTs.

These commercial DGA oil standards are currently provided in glass syringes of 30 ml. The shelf life of batches of gas-in-oil standards has been tested thoroughly and no significant changes in gas concentrations were observed over a period of several months. Figure 1, for example, indicates the DGA analyses performed over a six month period on a batch of gas-in-oil standards containing a prepared value of C2H4 of 97 ppm. The small, oscillating variations in gas content are related to the intra-lab reproducability of DGA analyses on the batch of oil standards over this long period of time. Similar results are observed for the other DGA gases.

Gas-in-oil standards are most useful for laboratories to determine the accuracy of their DGA results, and to improve it by detecting their hard-to-find biases. Accuracy should be checked each time a significant lab condition has changed (new operator, new equipment or a major piece of equipment like a gas extractor) and ideally at regular, frequent intervals. When outsourcing DGA analyses, gas-in-oil standards can also be used to verify if accuracy specifications are met by suppliers. This will ensure more accurate and reliable DGA and diagnoses.

References:

[1] M.Duval and Y.Giguere, Doble Conferences, 10C-01 (1984).
[2]. IEC Publication 60567, Clause 6.2 (1992).
[3]. ASTM Method D 3612-01, Annex A2 (2001).
[4]. M.Duval et al., Electra, 198, p.20 (2001).
[5]. R.Berube, Electricity Today, 8 (7), p.21 (1996].
[6]. Morgan Schaffer, www.morganschaffer.com.

M.Duval is with IREQ, Varennes, Canada. ET