Silicone Dielectric Fluids Fredi Jakob, Karl Jakob, Nicholas Perjanik Analytical ChemTech International, Inc. I. Introduction Mineral oil based dielectric fluids have been used more extensively than other dielectric fluids in electrical equipment because of their wide availability, low cost and excellent physical and electrical properties. Their only shortcoming is their relatively low flash and fire points. Polychlorinated bi-phenyls, PCB s, known generically as Askarels, were developed as alternative dielectric fluids. PCB s have excellent dielectric properties and they are far less flammable than mineral oils. Government agencies, at one time, mandated the use of PCB s whenever there was a safety concern related to fluid flammability. Unfortunately PCB s turned out to be an environmentally hazardous material. Federal regulations subsequently mandated elimination of PCB s. This regulation lead to a search for replacement dielectric fluids that are not as flammable as mineral oil. Two major types of fluids were developed. The first is a less flammable hydrocarbon fluid, which could be either a natural or synthetic blend of hydrocarbons. This type of hydrocarbon fluid is often designated as a high molecular weight hydrocarbon, HMWH. The second type of less flammable fluid type, which was first introduced in 1974, is a synthetic silicone based polymer. It has been estimated that as of 1993 over 100,000 transformers have been filled with silicone fluids and more than 25,000 have been filled with HMWH 1. The flash and fire points of these alternative fluids are compared with both mineral oil and Askarels in Table 1. 1
Table 1. Flash and Fire Points of Dielectric Fluids 1 Dielectric Fluid Askarel Mineral Oil HMWH Silicone Flash Point o C 195 150 284 300 Fire Point o C None 165 312 343 The large difference between the flash and fire points of silicone fluid, 43 0 C, compared to a difference of only 15 0 C for mineral oil is a significant safety factor. This difference is due to the low volatility of silicone fluids which results in very little vapor production when the fluid is heated. Complete combustion of silicone fluid results in the production of primarily silica, SiO 2, water and carbon dioxide. In the case of a fire the amorphous crust of silica that forms on the fluid surface reduces the heat flux to the burning fluid which in turn reduces the combustion rate. A primary fire concern is the heat release rate of a burning dielectric fluid. The greater the heat release rate the greater the danger of a transformer fire spreading to adjacent structures. Figure 1. Shows the heat release data for several types of dielectric fluids. 2
1600 Heat Release Rate (kw/m^2) 1400 1200 1000 800 600 400 200 Radiative Convective Total 0 Mineral Oil HMW Hydrocarbon Silicone Figure 1. Heat Release Data for Dielectric Fluids 1 II. Physical and Chemical Properties of Silicone Dielectric Fluids 2. Silicone fluids are synthetic linear polymers having the molecular structure shown in Figure H 3 C H 3 C H 3 C Si R O Si O Si R n CH 3 CH 3 CH 3 Figure 2. Molecular Structure of Synthetic Silicone Fluids 3
The organic groups, shown as R, can be virtually of any type and the R groups attached to any one silicon atom can be the same or different. The polymer terminating groups are generally trimethyls. The average value of n and the molecular weight of the attached organic groups determines the molecular weight of the polymer. Critical physical properties of the silicon fluid such as viscosity, volatility and thermal conductivity are determined by the molecular weight. Increased molecular weight results in increased viscosity, lower volatility and lower thermal conductivity. Silicone polymers in which both of the organic groups are methyl groups, CH 3, are called Polydimethysiloxanes (PDMS). The value of n in these fluids can vary from 0 to over 2000. Despite this very large range of molecular weights the PDMS fluids do not solidify over a wide range of temperatures. Viscosity increases from 0.65 to 2,500,000 centistokes as n goes from 0 to 2000. This wide range of viscosities allows the PDMS fluids to be used in a variety of applications such as dielectric fluids, anti-foam agents and heat- transfer fluids. These low toxicity silicone fluids are thermally and chemically stable and are also used, at low concentrations, in cosmetics, auto polishes and paints. A typical PDMS fluid, for use in transformers, in which all of the R groups are methyl groups has a viscosity of 50 centistokes. Table 2 provides comparative data for the three types of dielectric fluids discussed above. Values listed are based on the pertinent IEEE acceptance guides for each fluid 2. 4
Table 2. Physical and Chemical Characteristics of Dielectric Fluids,as Supplied 2 Property ASTM Method Mineral Oil HMWH Fluids Silicon Fluid Dielectric Strength (kv) D-877 30 25 35 Dissipation Factor (Power Factor) 25 o C 100 o C D-924 0.0004 0.009 0.0001 0.004 0.0001 0.0015 Pour Point o C D-97-40 -15-50 Specific Gravity 25 o C D-1298 0.875-0.910 0.869-0.910 0.957-0.964 Interfacial Tension (dyne/cm) 25 o C D-971 40 40 20.8 Viscosity 0 o C 25 o C 40 o C 100 o C D-445 D-2161 76.0 max. 14-16 12.0 max. 3.0 max. 2,200-2,900 350-379 120-140 13-19 81-92 47.5-52.5 35-39 15-17 Moisture Content, ppm max. D-1533 35 35 50 Flash Point o C D-92 150 238 (note) 268 (note) Fire Point o C D-92 160 311 (note) 371 (note) III. Maintenance of Silicone Fluids Dielectric fluids used in electrical equipment have three essential functions. They must be good insulators, provide thermal conduction and quench arcs. These essential functions must be maintained throughout the life of the fluid. Fluids used in transformers are all relatively stable and should last indefinitely in an ideal environment. In practice, dielectric fluids are subjected to thermal and electrical stresses that may alter the composition of the fluid. Undesirable materials may enter and mix with the fluids. Impurities, such as moisture and oxygen, can enter the equipment from the surrounding atmosphere or they can originate from materials, such as paper insulation and coatings, used in the construction of the equipment. Impurities interact chemically or physically with the 5
insulating fluids and alter their properties. Moisture from the atmosphere, or cellulose particles, from the paper insulation, dissolve or disperse in the fluid and reduce its insulating quality. Oxygen reacts chemically with mineral oil and can produce acid and sludge. Fortunately silicone fluids are more oxidation resistant than oil and acid formation is not a significant problem. The ASTM has developed a series of diagnostic tests for determination of the characteristics, physical properties, purity and functionality of insulating fluids. These ASTM methods were developed for evaluation of mineral oils but they are applicable, with minor modifications, to the evaluation of silicone fluids. The ASTM test methods that are recommended for maintenance testing of silicone fluids are listed in Table 3. 6
Table 3. Recommended Maintenance Tests for Silicone Transformer Fluid 3 Minimum Testing Test Procedure Acceptable result Unacceptable result may indicate Visual Inspection Crystal clear-free of particles Particulates, free water, color change Dielectric Breakdown ASTM D-877 >35 kv for fresh liquid >25 kv in transformer Particulates or water present Additional Recommended Testing Water Content ASTM D- 1533B Volume Resistivity ASTM D- 1159 < 100 Excess water present > 1 x 10 12 Water or contamination present Interfacial Tension ASTM D-971 ----- Contamination present Power Factor ASTM D-924 < 0-1% Polar/ionic contamination Viscosity ASTM D-445 50 + 5 cst Degradation or contamination Fire Point ASTM D-92 > 340 o C Contamination by volatile material Acid Number BCP ASTM D-974 ASTM D664 ----- Degradation of cellulose insulation or contamination Most of the ASTM methods listed in this table have been developed for the evaluation of mineral oils and can be used without modification for the evaluation of silicone fluids. For example, ASTM D-877 used to measure dielectric breakdown potential requires slight modification. This method calls for five consecutive measurements on the same oil sample. When it is applied to silicone fluids only one breakdown test can be made with each filling of the test cup because of the 7
buildup of arc decomposition products. The five dielectric breakdown measurements are therefore performed on five aliquots of the same sample.. IV. Fault Detection in Silicone Filled Transformers Dissolved gas analysis, DGA, has been universally accepted as a diagnostic tool for the detection of incipient faults in mineral oil filled transformers, load tap changers and other oil insulated equipment. Application of the DGA method to silicone fluid filled electrical equipment has generated much interest. The development of the DGA method for silicone fluids parallels that for application of DGA to mineral oil filled transformers. Griffin 4 completed extensive laboratory studies on the degradation of silicone fluids subjected to thermal and electrical stresses. He studied the conditions that are associated with the common fault conditions viz. overheating, corona or partial discharge and arcing. The low molecular weight gases that were found are identical to those that are normally reported, with DGA, for mineral oils: hydrogen, carbon monoxide, carbon dioxide, methane, ethane, ethylene and acetylene. The concentration of each of these fault gases depends on many parameters including temperature, fault energy, fault characteristics, oxygen level and the presence of metal catalysts such as copper. Several fault gas patterns were reported by Griffin. Thermal faults produce mainly methane with small amounts of hydrogen, carbon monoxide, ethylene and ethane. Corona produce mainly hydrogen and methane. Arcing results in the production of hydrogen, methane, carbon monoxide and acetylene. The ratio of hydrogen to acetylene is much higher in silicon fluids than it is in mineral oils subjected to the same arcing conditions. In addition to variations in the dissolved gas concentrations, the concentration of the gases above the fluid will also vary. The Ostwald equilibrium solubility of fault gases in mineral oil is 8
different than it is in silicon fluids. This difference in solubility, which is tabulated in Table 4, will lead to different equilibrium concentrations of fault gases in the transformer fluid and gas blanket. Remember that these are equilibrium values that provide limiting concentration ratios and they may not be applicable to the dynamic environment inside of an operating transformer. Table 4. Ostwald Solubility Coefficients 5 GAS Silicone Fluid Mineral Oil Hydrogen (H 2 ) 0.057 0.0429 Nitrogen (N 2 ) 0.143 0.0745 Carbon Monoxide (CO) 0.096 0.102 Oxygen (O 2 ) 0.175 0.138 Methane (CH 4 ) 0.514 0.337 Carbon Dioxide (CO 2 ) 1.401 0.9 Acetylene (C 2 H 2 ) 1.411 0.938 Ethylene (C 2 H 4 ) 1.018 1.35 Ethane (C 2 H 6 ) 1.339 1.99 Interpretation of DGA data is empirical in nature and the most widely used empirical methods, for mineral oil filled transformers, are based on the identification of key gases and various gas concentration ratios. Threshold concentrations for each of the key gases and for the 9
total fault gas concentration must also be empirically established. Correlation of key gases and investigated faults are essential to the development of these methods. At this time the data available for mineral oil filled transformers is far greater than that available for silicone filled units. General guidelines for the interpretation of gases generated in silicone immersed transformers have been developed by IEEE 5. The gases associated with fault types are essentially the same as those observed by Griffin in his laboratory studies. Cellulose decomposition results in the production of both carbon monoxide and carbon dioxide in a concentration ratio that is temperature dependent. The presence of carbon oxides, in the absence of hydrocarbons, is a significant indication of cellulose decomposition. Determination of dissolved furanic compounds in insulating fluid, according to ASTM D-5837, is recommended as a confirmatory test to assess the extent of cellulose degradation. IEEE guidelines recommend the establishment of base line data when the transformer is first placed in service. A weekly or monthly DGA analysis is recommended when a new transformer is initially placed in service. The rate of gas generation is a key parameter to indicate the severity of an incipient fault. At this time no specific gas generation rates have been presented in the IEEE guidelines but the presence of any amount of acetylene is considered extremely serious. For existing units, where no base line data is available, IEEE recommends a set of threshold concentrations based on collective laboratory experience 5. These flag points are summarized in Table 5. If any of the individual fault gases or the total dissolved combustible gas, TDCG, exceeds the flag points one should take additional samples to assess gas generation rate and the severity of the fault. 10
Table 5. IEEE Dissolved Gas Threshold Levels in ppm vol 5 GAS H 2 CH 4 C 2 H 2 C 2 H 4 C 2 H 6 CO CO 2 TDCG Threshold Level 200 100 1 30 30 3000 30000 3360 The second stage in the interpretive process involves interpretation of the fault gas data. Current IEEE recommended interpretation rules are summarized in Table 6. As more silicone DGA data becomes available these rules may be modified or expanded. Table 6. Key Gas Interpretation Fault Type Key Gas(es) Additional Gases Notes Thermal-Fluid CO(principal), CO 2, CH 4 H 2 and C 2 H 6 1 Thermal-Cellulose CO(principal) and CO 2 ----- ----- Corona H 2 (principal) and CH 4 CO, C 2 H 2, C 2 H 4, C 2 H 6 ----- Arcing H 2 (principal) and C 2 H 2 CH 4, C 2 H 4, CO, CO 2 2 1. Oxygen must be present to form CO and CO 2. Methane, CH 4, is favored with low oxygen levels. 2. The ratio C 2 H 2 /H 2 increases with increasing arc energy. Ratio methods similar to the widely accepted Rogers Ratio method are not yet available in the IEEE guide. Ratios that may be significant might include hydrogen/acetylene and carbon 11
dioxide/carbon monoxide. Case histories for DGA applied to silicone fluids are limited. It is interesting to take a case history which is presented in the IEEE guideline 5, and analyze the data with the DTK plus software package. The silicone filled 3500 KVA transformer in question was initially tested after being in service for three years. None of the individual fault gases or the total dissolved gases exceeded the IEEE concentration threshold levels, given in Table 5. The transformer was therefore considered to be operating normally. Thirteen months later every key fault gas and the TDCG exceeded the IEEE threshold values. Inspection revealed arcing in one of the windings. This wind was rewound and a catastrophic failure was avoided. Appendix A shows the plots and report generated with DTK plus for this interesting case history. The procedures for sampling silicone fluid filled transformers are identical to those employed for mineral oil filled units. The sampling procedures are documented in ASTM D- 3613. The determination of the fault gases, and oxygen and nitrogen in the oil sample is conducted according to ASTM D-3612. The gases that are separated and quantified by this ASTM method include nitrogen, oxygen, hydrogen, carbon monoxide, carbon dioxide, methane, ethane, ethylene and acetylene. V. Conclusion Silicone fluids are a viable alternative to polychlorinated biphenyls. Their high flash and flame points make them an attractive choice when safety considerations mandate a less flammable dielectric fluid. These fluids are non-toxic and generate only one toxic substance, carbon monoxide, when they are ignited. Silicone fluids are oxidation resistant and thus acid buildup, which is a problem with mineral oils, is not a significant problem with silicone fluids. 12
Silicone fluids have higher viscosities than transformer grade mineral oils and thus their overall heat transfer rates are lower than that of mineral oils. In the worst case scenario the rated temperature rise will be achieved with 90 percent of the name plate rating. Thus, transformers that will be run at nameplate rating would not be suitable candidates for retrofilling with silicone fluids. VI. References 1. Goudie, J.L., "Less Flammable Transformer Liquids Gain Acceptance", CEE News, Feb. 1993. 2. IEEE C57.111-1984, "IEEE Guide for Acceptance of Silicone Fluids and its Maintenance in Transformers". IEEE C57.121-1988, "IEEE Guide for Acceptance and Maintenance of Less Flammable Hydrocarbon Fluid in Transformers". IEEE C57.106-1991, "IEEE Guide for Acceptance and Maintenance of Insulating Oil in Equipment. 3. 1996 Annual Book of ASTM Standards, Vol. 10.03, "Standard Test Methods for Silicone Fluids Used for Electrical Insulation", Designation: D2225-92, American Society for Testing and Materials, Philadelphia, PA 1996. 4. Griffin, P.J., "Analysis for Combustible Gases in Transformer Silicone Fluids," Minutes of the Fifty-Second Annual International Conference of Doble Clients, 1987, Section 10-701. 5. IEEE Trial-Use Guide for the Interpretation of Gases Generated in Silicone-Immersed Transformers," IEEE Draft P1258, IEEE, NY, NY September 1995. 13