Impact of a Fire in Electrical Equipment Containing Insulating Oil Contaminated with Polychlorinated Biphenyls

Transcription

1 21 rue d Artois, F PARIS C3-101 CIGRE Impact of a Fire in Electrical Equipment Containing Insulating Oil Contaminated with Polychlorinated Biphenyls M-C. LESSARD, J. CASTONGUAY, B. LAROSE, J. BOYD, S. FORTIN, M. PLANTE Hydro-Québec Canada SUMMARY A fire in a power transformer containing some oil contaminated with polychlorinated biphenyls (PCBs) may have a significant socio-economic impact. It would result in the partial combustion of the oil and thus generate a large quantity of smoke and soot. The presence of PCBs in the insulating oil would also lead to the creation of polychlorinated dibenzofurans (PCDF) and dibenzodioxins (PCDD), two chemical compound families better known as dioxins and furans and known to be highly toxic, that would be adsorbed in the soot. The impact of the event would create a risk of contamination for the environment as well as for the workers and people in the path of the fire plume. However, the production rate of toxic contaminants has never been assessed in uncontrolled combustion. To compensate for this lack of data and find quantitative answers aimed at assessing the environmental, health and safety risks as well as facilitate firefighting operations, Hydro-Québec (HQ) conducted combustion tests on mineral insulating oil that was highly contaminated with PCBs (2,000, 10,000, 15,000 and 20,000 mg/kg (ppm)). The tests were used to establish a simple combustion model which served to determine: A quantitative measurement of the soot that was produced; An analysis of the content of residual PCBs in the soot as well as of toxic congeners (coplanar PBCs), polycyclic aromatic hydrocarbons (PAH) and dioxins (PCDD) and furans (PCDF); An assessment of the respective production rate of these products by comparing their relative toxicity. The chosen approach consists in comparing the impact of a fire involving mineral oil not containing any PCBs with the impact of a fire involving mineral oil containing a known concentration of PCBs. Based on the analytical results, certain conclusions could be drawn: A fire involving mineral oil without any PCBs will produce soot, carbon monoxide (CO) and PAHs (with the latter by-product presenting the greatest risk of exposure); An oil fire involving PCBs will generate the same by-products as oil without any PCBs but with a proportion of PCDD, PCDF, residual PCBs and coplanar PCBs; The creation of PCDD and PCDF is proportional to the concentration of PCBs in oil; lessard.marie-claude@ireq.ca

2 The quantities of PCDD and PCDF produced during combustion are extremely small. Two factors account for these results: first, when there is a fire, a very small proportion of PCBs are transformed into PCDD and PCDF since most of the PCBs are burned and, second, PCB levels in the contaminated oils are small (2% in the case of oil contaminated at a level of 20,000 mg/kg). This study clearly shows that it is soot and smoke containing products such as PAHs emitted during such a fire that are primarily responsible for the toxicity of air emissions. In fact, based on the toxic risk criteria generally accepted by the scientific community, the production of toxic compounds as a result of PCBs in oil would significantly contribute to the toxicity of soot only at PCB levels greater than 20,000 mg/kg in oil. In the 1990s, HQ voluntarily withdrew a large proportion of its PCB-contaminated equipment and oils. In 2008, the Canadian government adopted a regulation on PCBs that requires electrical equipment with PCB levels greater than 500 mg/kg to be withdrawn from service by the end of 2009 with a possible extension until These two initiatives now allow the utility to consider all electrical equipment fires as common-hydrocarbon combustion. Hence, PCB and chlorinated PCDD and PCDF levels in soot and smoke do not represent an additional toxic risk in relation to the actual soot during a fire involving PCB-contaminated oils containing less than 20,000 mg/kg of PCBs. The myth of highly toxic fumes being produced from fire involving slightly or moderately PCB-contaminated oil therefore does not hold ground and was clarified in this study. KEYWORDS Polychlorinated Biphenyls-PCBs Fire - Mineral insulating oil Furans PCDF Dioxins PCDD - Environmental contamination - Health and safety risks - Electrical equipment 1. INTRODUCTION A fire in electrical equipment containing a large amount of mineral oil, such as a power transformer, requires and will always require an efficient response to limit the risk of fire propagation and exposure to people in the surrounding area. However, at present, if the insulating oil in the equipment is contaminated with polychlorinated biphenyls (PCBs), the impact of the incident is increased tenfold. In addition to generating a large quantity of smoke and soot, the partial combustion of the PCBcontaminated mineral oil will release a wide range of toxic by-products such as residual PCBs (not burned), polychlorinated dibenzofurans (PCDF), and polychlorinated dibenzodioxins (PCDD), commonly known as dioxins and furans. In the past, Hydro-Québec (HQ) had to deal with several fires involving PCBs and the resulting media coverage. Table 1 shows the main incidents. As reference, an environmental assessment conducted during the 1989 incident roughly established the air emissions (analyzed with the knowledge and technologies available at the time) of the three main types of toxic compounds: PCBs: 18 g, PCDF < g and PCDD < g. These relatively small quantities were already considered negligible at that time. Even though this is a topic of considerable interest for electrical utilities, very few studies have directly covered this issue [1][2] and mainly involve assessing the formation of such toxic compounds through the thermal degradation of PCBs [3]. Erikson et al. [4][5] have thus established that the generation of PCDFs was at a maximum level during controlled-oxidation tests on PCBs and mineral oil at 675 C. These studies also showed that trichlorobenzene (TCB), a solvent used in combination with certain Aroclors, would also produce PCDDs and PCDFs under similar conditions. However, the experimental systems used at the time did not seem to generate soot as normally observed during a 2

3 fire in electrical equipment insulated with mineral oil. Note that since the late 1980s, the start of the elimination of PCBs worldwide, research work on PCB combustion virtually stopped. Table 1: Fires in facilities involving PCBs Year Event 1984 A fire in a capacitor bank insulated with PCBs at IREQ s Highvoltage Laboratory required a major and costly decontamination of the premises and close medical monitoring of the responders involved during the fire A fire in a power transformer at Longue-Pointe substation resulted in the combustion of over 25,000 litres of PCB-contaminated oil (144 ppm of Aroclor -1260) and the evacuation of residents living near the substation A fire in a transformer containing PCB-contaminated oil (84 ppm of Aroclor -1260) at Bout-de-l Île substation resulted in the evacuation of hundreds of Montreal East residents. In a full combustion, the fuel-oxidant (oxygen) balance is perfect and the flame is very hot. In the case of oil [8][9][10][11], the long hydrocarbon molecules are more difficult to break into small fragments which, by oxidizing, supply energy to the flame. In this environment, the fuel content is too high (polyacetylene and other heavy radicals) such that oxidation does not occur and the surplus of reactive species condensated (into more thermodynamically stable ring structures) to form PAHs and ultimately soot. The latter are in fact made of up virtually fully dehydrogenated carbon cycles. In the case of combustion of the same oil but with different PCB levels, three other types of toxic and highly toxic by-products are produced, i.e. residual PCBs, coplanar congeners PBCs, PCDDs and PCDFs. They result, as predicted, from the partial destruction of PCBs where they are formed by the partial oxidation of PCBs and TCBs. A simplified diagram of these mechanisms is presented in Figure 2. Mineral oil Mineral oil with PCBs CO 2 Soot PAHs H 2 O CO PCDFs and PCDFs Residual PCBs Coplanar PCBs Figure 2: Combustion mechanism of mineral oil and mineral oil containing PCBs Starting in 1980, HQ stopped purchasing equipment insulated with pure PCBs. The utility implemented its PCB management and elimination program in 1985 despite the fact that there were no regulations on the use of PCBs in electrical equipment. In 1998, over 52,000 pieces of equipment containing pure PCBs were withdrawn from service and destroyed. In 2008, a new Canadian regulation on PCBs was adopted which stipulated withdrawing in-service equipment containing over 500 mg/kg of PCBs before December 31, 2009 (possible extension to December 31, 2014). In 3

4 addition, based on a recent survey of the utility s data banks, the elimination of PCB-contaminated oil with a concentration of 50 mg/kg is almost complete (99.6% of the 2,931 pieces of equipment). Note that the procedure was done strictly for equipment with a large volume of oil, i.e. with a rating of over 49 kv, since this type of apparatus presents the highest risk of toxicity in the event of a fire. HQ thus conducted a study to determine the real environmental and toxic impacts of a fire involving mineral oil with small levels of PCBs. Laboratory work [6][7] required the development of a specific and safe test bench capable of simulating the behaviour of a fire involving mineral oil leaking from a transformer and potentially contaminated with PCBs. This reduced-scale test plan was designed to entirely confine the soot that was generated, enable it to be quantitatively measured and recovered for chemical analysis. 2. EXPERIMENTAL DESIGN The chosen approach allowed the effects of a fire involving naphthenic insulating mineral oil (Volt Esso 35) free of PCBs (0 mg/kg (ppm)) to be compared against a fire involving the same oil containing 2,000, 10,000, 15,000 and 20,000 mg/kg (ppm) of PCBs (Aroclor -1260). Preliminary tests were done with oil that had been previously contaminated with PCBs (100 and 500 ppm). The final test bench (baffle chimney), developed in HQ s laboratories, is presented in Figure C Condenser Filter Chimney made of trays 140 C Oil cup (7.5 cm ø) Figure 1: Test bench allowing the combustion of oil and recovery of generated soot Following each combustion test (5 ml of oil), the soot was recovered on the various trays (removable individually). The filter would trap slightly more than half the soot, with the remainder being deposited on the four lower trays. The tests were replicated (5) to validate the repeatability of the test procedure and setup. In addition to the analysis of the residual PCBs, PCDDs and PCDFs, the toxic congener (coplanar PBCs) and PAH levels were measured. These analyses were all done by an outside laboratory accredited by Environment Canada. Note that PCDDs and PCDFs are found in a very large number of compounds (congeners). Only the 17 members of these two families with a symmetrically tetra-substituted basic structure were considered, since these have the highest toxicity levels. Relative toxicity factors were used to establish 4

5 the Toxic Equivalency Factor (TEF) [12] of the PCDD-PCDF mixtures measured in the soot. Their relative toxicity was standardized into a single value to simplify the interpretation of the results. 3. RESULTS 3.1 Characterization of soot The analysis results for each combustion test were compiled by type of toxic product. Figure 3 shows the changes in each based on the PCB content in the oil during the test. Each replica was considered. Minimum and maximum curves were used, rather than a mean curve, to show the reproducibility of the tests and dispersion around this mean. The resulting dispersion (standard deviation below 13% for each by-product) appeared to be highly acceptable given the complexity of the procedure. The tests showed that regardless of the concentration of PCB in oil, about 17% of the burnt mass of oil is found in the form of soot after combustion (Figure 3). PAH production decreases very slightly when PCB levels increase. Conversely, residual PCB levels in the soot are proportional to the initial PCB levels in the oil. Given that soot represents about 17% of the mass of oil that was burned and that the PCB levels in the soot correspond to about 50% of the concentration in oil, it can be deduced that only 8% of the residual PCBs remain, thus confirming that over 92% of the initial PCBs are destroyed by the flames. PCDF and PCDD concentrations in soot are also proportional to the initial PCB levels in oil, but at a much lower concentration than for the residual PCBs. Note also that PCDFs are found to a much greater extent than PCDDs, i.e. 78 times more. Their formation is most likely favoured by the partial and direct oxidation reaction of the PCB molecules, which is substantially favoured by the PCBs very structure. 3.2 Modeling of oil combustion rate and toxic product formation A simple mathematical model was established based on the results (maximum values) of the combustion of contaminated oils shown in Figure 3. The model is used to quantitatively establish the variation in the formation rates of the various toxic by-products based on the PCB levels in contaminated mineral oil. These are presented in Table 2. To simplify the model, the basic parameters for the calculation were chosen to define a typical oil fire of 10 m 2 (diameter = 3.57 m) generating soot corresponding to 20% of the mass of burnt oil at a rate of 1.25 L/m 2 /min. 5

6 Figure 3: Characterization of soot: variation in the formation rates of the various by-products based on PCB levels (replicas=5) Table 2: Results of the combustion of oil highly contaminated by Aroclor PCBs for a burning surface of 10 m 2 Generated products Production factor Quantities produced (PCB level= 10,000 ppm) Quantities produced (PCB level= 20,000 ppm) Quantities produced (units) Oil mass (kg) kg/min Volume of oil (L) L/min Soot (kg) 20% kg/min PAH (mg) 2.7% mg/min Residual PCBs (mg) (*) mg/min Toxic PCBs (TEF mg) 7 x 10-5 (*) mg/min PCDD-PCDF (TEF mg) 7 x 10-5 (*) mg/min Σ Toxic chlorinated mg/min compounds (TEF mg) * Production factor per ppm of PCB/oil: slope of Figure 3 curves 6

7 Given the relative quantities that were produced, the contribution of the PCB coplanar congeners (toxic PCBs) to the toxicity of all the combustion products is very similar to that of the PCDDs- PCDFs, while the PAH contribution is substantially higher. 3.3 Relative toxicity level of the toxic products generated To compare the relative toxicity of each of the main toxic compounds formed by the combustion of contaminated oil, the acceptable maximum concentration for each compound was estimated based on the main effects to be prevented in each compound for a six-hour exposure without any significant health risk. These limits are listed in Table 3 and are used to determine the relative importance of the potential exposure to toxic compounds in relation to this study. Note that these four families of toxic substances (i.e. PAH, coplanar PCBs, PCDDs and PCDFs) are basically adsorbed in soot rather than released in combustion gases (H 2 0, CO and CO 2 ). These results show that the PAH require greater dilution to attain the acceptable exposure limit, followed by soot particles and toxic chlorinated compounds (PCB and PCDD-PCDF coplanar congeners). However, in the latter case, this level of toxic risk is only attained with a concentration of 20,000 mg/kg of PCBs in oil; for lesser levels, it will be reduced proportionately. The greatest toxic risk of soot and smoke is closely linked to their physical characteristics: a substantial number of very small particles with large adsorption surfaces. Once inhaled, they penetrate deeply into the lungs, which makes them all the more difficult to get rid of. Table 3: Potential exposure associated with oil highly contaminated with PCBs (20,000 ppm Aroclor -1260) Toxic products Acceptable exposure limit (mg/m 3 ) Dilution required to comply with standards (volume in m 3 ) (for 1 g of generated soot) Relative potential for exposure Soot (kg) 3.5 1, PAH (mg) , Residual PCBs (mg) Toxic PCBs (TEF mg) 1 x 10-5 (*) PCDD+PCDF (TEF mg) 1 x 10-5 (*) Σ Toxic chlorinated compounds 1 x 10-5 (*) (TEF mg) CO (mg) 40, * Values estimated by HQ for a 6-hour exposure. 4. CONCLUSIONS A fire involving mineral oil without any PCBs will always generate toxic products such as soot, CO and PAHs. The latter present the greatest potential for exposure in this type of fire [13]. A fire involving PCBs will generate the same products, but also a certain quantity of PCDDs, PCDFs, and residual PCBs, including toxic PCBs. This study has shown that the formation of the latter is proportional to the oil s concentration of PCB contamination. It was also noted that the PCDD and PCDF quantities generated during combustion are extremely small. Two factors account for this: first, in a fire, there is little transformation of PCBs into PCDDs and PCDFs since most of the PCBs are burned. Indeed, the PCB content in the contaminated oils is relatively low (an oil with a contamination level of 20,000 mg/kg contains approximately 98% oil and only 2% PCBs). The main tests and analyses conducted for this study enabled the main toxic products formed in both cases (with and without PCB contamination) to be identified and quantified. Their relative toxicity 7

8 levels could thus be evaluated. When combined with the measurement of their respective formation rates, it was possible to conclude that a fire will generate by-products (smoke and soot) with a toxicity level greater than that of a fire involving pure mineral oil containing 20,000 mg/kg (ppm) of PCBs or more. Based on the conclusions of this study and considering that HQ s equipment contains almost no PCBcontaminated oil at concentrations greater than 50 mg/kg, it will be clearly possible to deal with a fire in equipment containing oil contaminated with PCBs at a level of up to 20,000 mg/kg as a mere hydrocarbon fire. Therefore, it will not be necessary to take measures in addition to those used in traditional firefighting procedures. If an evacuation of local residents is deemed necessary, it will have to be justified solely on the basis of the magnitude of the fire, unfavourable winds, the direction of the smoke, and the risks presented by the products generated during any usual combustion, while taking into account the exposed environment and not the fact of whether or not there may be any PCBs in the oil. BIBLIOGRAPHY [1] D. Train, A. Chamberlant, D. Dupont and J. Castonguay, Incendie des condensateurs au BPC au Laboratoire Haute tension. IREQ et la décontamination subséquente, Internal Report No. IREQ-5RT3608G, January [2] Fires in Electrical Equipment Contaning PCBs: Recommendations to Prevent Contamination by PCDFs, Environmental Protection Service, Environment Canada, Report EPS 1/CC/2, March [3] S.E. Swanson, M.D. Erikson and L. Moody, Products of Thermal Degradation of Dielectric Fluids, US-EPA Publication No. 560/ , September [4] M.D. Erikson, C. J. Cole, J.D. Flora, G.G. Gorman, C.L. Haile, F.C. Hopkins, S.E. Swanson and D.T. Heggem, Thermal Degardation Products from Dielectrics Fluids, US Environnmental Proctection Agency, Office of Toxic Substances, Washington, EPA-560/ , [5] M.D. Erikson, C. J. Cole, J.D. Flora, G.G. Gorman, C.L. Haile, F.C. Hopkins, S.E. Swanson and D.T. Heggem, PCDF Formation from PCBs under Fire Conditions, Chemosphere, Vol. 14, No. 6-7, 1985, pages [6] J. Castonguay, Combustion d huile minérale contaminée aux BPC: Phase 1, Étude de la production de suies et de leur toxicité (contenu en BPC, PCDD et PCDF), Internal Report No. IREQ C, October [7] J. Castonguay, Combustion d huile minérale contaminée aux BPC: Phase 2 et 3, Étude de la production de suies et de leur toxicité (teneur en HAP, BPC, PCDD et PCDF). Internal report No. IREQ C, November [8] C.P. Bankston, B.T. Zinn, R.F. Browner and E.A Powell, Aspects of the Mechanism of Smoke Generation by Burning Materials, Combustion & Flame, Vol. 41, 1981, pages [9] R.L. Dod,, N.J. Brown, F.W. Mowrer, T. Novakov and R.B. Williamson, Smoke Emission Factors from Medium-Scale Fires: Part 2, Aerosol Sci. & Tech., Vol. 10, 1989, pages [10] U.O. Koylu and G.M. Faeth, Carbon Monoxide and Soot Emissions from Liquid-Fueled Buoyant Turbulent Diffusion Flames, Combustion and Flame, Vol. 87, 1991, pages [11] U.O. Koylu, Y.R. Sivathanuet and G.M. Faeth, Carbon Monoxide and Soot Emissions from Liquid-Fueled Buoyant Turbulent Diffusion Flames, Fire Safety Science: Proceedings of the Third International Symposium, 1992, pages [12] Committee on EPA's Exposure and Human Health Reassessment of TCDD and Related Compounds, National Research Council, Health Risks from Dioxin and Related Compounds: Evaluation of the EPA Reassessment, The National Academies Press, Washington, 2006, page 35 et pages [13] M. D. Hays, L. Beck, P. Barfield, R. D. Willis, M. S. Landis and R.K. Stevens, Physical and Chemical Characterization of Residual Oil-Fired Power Plant Emissions, Energy & Fuels, Vol. 23, 2009, pages