Dissolved gas analysis results from the online monitor, showing gas levels recorded over two years following energization. The dashed lines indicate when the monitor was offline. 

Context in source publication

Context 1

... on the basis of laboratory tests [5], [8], [13], [14]. The temperatures of the oil and the paper insulation were continually monitored, because such data could be useful in in- terpreting any observed changes in the condition of the insulation over long periods, or a phenomenon such as unusually heavy gassing. The highest temperatures recorded by the array of fiber-optic probes within the transformer are shown in Figure 3; they do not suggest overheating of paper or oil. The water content of the oil was calculated from the output of one of the Vaisala probes. At room temperature the solubilities of water in mineral oil and in vegetable oil are approximately 50 and 1,100 ppm, respectively, so that a higher water content is to be expected in a transformer using vegetable oil. The water concentrations calculated from the Vaisala probe data (Figure 4) agree fairly well with the Karl Fischer titration measurements shown in Table 2. On energization of the transformer the water content of the oil was 30 ppm. It fell to 22 ppm after 5 months of operation, and the same value was observed after 25 months. The difference between oil water content measured by Karl Fischer titration, and that calculated using the Vaisala probe, may be due to absorption of water by the oil sample in transit from the transformer to the laboratory. The insulation at the top of an energized transformer is usually hotter than the insulation at the bottom, and therefore water released from the warmer cellulose at the top will tend to be adsorbed on the cooler cellulose at the bottom. Over time this adsorbed water will slowly migrate toward the center of the cellulose. Du [15] calculated the time taken for water to diffuse through 1-mm-thick pressboard impregnated with mineral oil as 333 hours ( ≈ 14 days) at 20°C. Because much thicker blocks of cellulose had been used in the construction of the transformer, the corresponding diffusion times would be much longer. Consequently, the water content of the oil may change slowly over a long period, as water is exchanged between the center of the thick cellulose block and the bulk oil. Toward the end of its useful life mineral oil forms a sludge, whereas vegetable oil becomes more viscous. The thermal performance of a fluid is related to its viscosity [16], [17]. It may therefore be possible to assess the condition of the vegetable oil by monitoring the temperatures within the transformer and cor- relating them with the load, i.e., when the viscosity changes the temperature distribution may change sufficiently to be detected. If the temperatures deviate from those expected, the usual oxidation tests, which may include measurement of oxygen inhibitor content, dielectric dissipation factor, and acidity, should be carried out. Fiber-optic probes were inserted into different sections of the windings to locate the hottest points within the transformer. Ex- cessive heat and oxygen can together degrade oil and cellulose. Consequently, it is necessary to ensure that the cooling of the transformer is adequate and the number of hot spots is mini- mized. Oil FR3 is known to produce higher levels of ethane (C 2 H 6 ) and hydrogen than mineral oil under nonfault conditions [5], [7], [8], [13]. Some of the components of soybean vegetable oil, such as linolenic acid, generate ethane by reacting with oxygen [18]. A catalyst, e.g., copper, is required for the ethane-pro- ducing reaction to proceed. The reaction noted by Schaich [18] may be the source of ethane production within the transformer. Atanasova-Hoehlein et al. [19] suggest that ethane is generated by the lipid peroxidation mechanism, which can occur in all omega-3 unsaturated fatty acids. They also suggest that ethane can be considered as the main gas involved in thermal-oxidative degradation of vegetable oils. Mineral oil generates less ethane than does vegetable oil because of differences in hydrocarbon molecular structure, i.e., ring structures in mineral oil but straight chains in vegetable oil triglycerides. Consequently, ethane generation within a transformer may be related to the proportion of linolenic acid form- ing the triglyceride, the temperature, the availability of oxygen, and the copper surface area exposed to the oil. A common measure of gas solubility is the Ostwald coefficient, which is the concentration of gas dissolved in the oil divided by the concentration of free gas in the headspace of a vessel, such as a sampling syringe [20]. Thus the concentration of a gas dissolved in the oil can be calculated from a measurement of the concentration of the same gas in the headspace of the syringe. The Ostwald solubility coefficients for various gases in FR3 and in mineral oil are given in Table 3 [21], [22]. The levels of dissolved gas in the transformer oil were monitored for nearly two years. The ethane level increased around the time of energization, plateaued at approximately 120 ppm, and remained at that level for nearly two years (Figure 5). It would appear that the ethane-generating reactions slowed down and possibly ceased. The concentration of hydrogen fell, perhaps because hydrogen was consumed in further reactions. The online dissolved gas analysis measurements agreed satisfactorily with the independent laboratory measurements (Table 4). The dissolved gas content of the oil in the stored transformer was measured (Table 5) to compare its dissolved gas analysis signature with that of the operating transformer. No ethane was detected during factory acceptance tests conducted in July/Au- gust 2008. However, two years later ethane was found in concentrations comparable with those in the operating transformer. The second transformer was energized only during carefully con- trolled factory testing two years prior to the sampling; therefore, it would appear that ethane can be produced in the absence of a fault, in agreement with Duval’s observations of stray gassing [7], [8]. High ambient temperatures may have been responsible. A major advantage of using a vegetable oil is that, if a leak occurs, the oil will be consumed by microorganisms. The manu- facturers of FR3 noted some speculation that natural ester insulation fluid may support microbiological growth in transformers; however, their eight-year study did not produce any supporting evidence [23]. The food industry has carried out much research on spoilage prevention [22]–[26]. One method is to limit access to water, thus preventing the growth of microorganisms. The term “water activity” was first used by the food industry to determine the effect of the water content of a food on its spoilage [25] and is now used in connection with loss or gain of water by a food in a given environment [26]. It is a ratio, based on water vapor pressure, and covers the range 0 to 1, where 0 = dry and 1 = saturation. The minimum water activity levels required to sustain various organisms are given in Table 6 [24]. It is assumed in the water activity approach that the system is in thermodynamic equilibrium, contrary to the usual situa- tion in transformers. However, it may be reasonable to assume that, provided the ratio (instantaneous water vapor pressure/ maximum water vapor pressure at the same temperature) is kept below the relevant water activity, organisms will not survive within the transformer tank. The solubility of vegetable oil in water is around 1,000 ppm at room temperature and increases with increasing temperature. The standard ASTM D6871 Standard Specification for Natural Ester Fluids Used in Electrical Apparatus [27] specifies a maximum oil water content of 200 ppm (the breakdown voltage of FR3 falls at around 300 ppm [1]). Thus, if the oil water content is kept below the level speci- fied by the ASTM standard, the oil would be expected to be too dry for microorganisms to survive within the transformer tank and degrade the oil. Regular monitoring of the condition of an oil allows a vari- ety of problems to be detected and rectified before the overall operation of the transformer is affected. However, without full lifetime data it can be difficult to establish the significance of a given parameter value for the condition of a transformer. Few data are available for vegetable-oil-filled transformers. The dielectric dissipation factor (DDF) of an oil is a function of its relative permittivity and conductivity, both of which are normally higher for a vegetable oil than for a mineral oil. It is expected that the DDF and acidity of an oil (vegetable or mineral) will increase as the oil ages. Work is continuing to predict the likely effect on the insulation of a transformer of the compounds that cause its DDF to increase, e.g., acids [28]. The DDF of the FR3 in the two transformers was measured in our laboratory, at various temperatures between ambient and 90°C, following IEC 61620 [29], and in a commercial laboratory, at ambient temperature and at 90°C, following IEC 60247 [30]. Figure 6 shows the changes in DDF over the first two years of operation. In Table 7 our measurements are compared with those made by a commercial laboratory. It can be seen that there is reasonable agreement between the two. Some differences between the DDF values for the in-service transformer and the stored transformer can be seen in Figure 6, but they are small relative to the maximum value 0.005 suggested for new vegetable oil in the IEEE Guide for Acceptance and Maintenance of Natural Ester Fluids in Transformers [31]. In this standard the suggested DDF limit (0.005) is applicable only to the natural ester in new equipment; at the time of writing, insufficient data were available to allow specification of limits for service-aged oil. However, prompt investigation is recom- mended in the IEEE guide if the dissipation factor exceeds 0.03 at 25°C. Another standard, developed for synthetic organic esters, recommends a maximum value of 0.01 at ambient temperature [32]. Although the DDF values for the two ...