Technical report No. 1
Dr Carl Meyer, Dr Tom Beer and Dr Jochen Müller
Department of the Environment and Heritage, May 2004
ISBN 0 642 54993 1
The review, Sources of Dioxins and Furans in Australia: Air Emissions 1998, previously estimated the total emission of dioxins and furans to air from prescribed burns and bushfires to be 72 to 1,700 g TEQ per year. Using the measured field emission rates from the current study the total emission of dioxins, furans and dioxin-like PCBs to air is estimated to be 31-495 g TEQ per year, 70% lower than the previous estimate.
Based on overseas literature, biomass burning from bushfires, prescribed fires, and agricultural stubble fires was thought to be the major source of toxic dioxin, furan and coplanar PCB emissions in Australia. However, because there were no direct measurements of emissions from Australian fuels the uncertainty was very large, with a twenty-fold range from the minimum to the maximum. The bushfire project was commissioned to reduce this uncertainty by directly measuring the emissions of dioxins, furans, and coplanar PCBs from fires in the field, or in laboratory environments that adequately reproduce field combustion characteristics. These data were then used to reassess the emissions from field burning of biomass in Australia during the last decade.
The project found that dioxin and furan emissions in Australia may have been previously overestimated by at least a factor of three. Measured emission rates from fires in the field were at the minimum of the previous estimates, because these previous estimates were largely based on laboratory measurements. The discrepancy arises due to differences in combustion characteristics between field fires and laboratory simulations.
The project was designed to measure dioxin emissions from 21 field burns and 19 laboratory tests so as to replicate closely the combustion processes of open fires in the field. Field sampling requires the ability to sample very large volumes of smoke within the short duration of many field fires, typically from 15 minutes to 3 hours. A set of high volume samplers were designed and constructed that could be operated from the tray of a 4WD vehicle. These units could be quickly relocated to stay within the smoke plume as the fire front progressed. The samplers also measured total suspended particulate matter (TSP) and carbon dioxide (CO2) concentration in order to relate the dioxin concentrations measured in the smoke plume to the combusted fuel mass, and thus, to derive emission factors. The field burns comprised 13 prescribed fuel reduction fires in SE Queensland, Central Victoria, and SW Western Australia, 2 cane burns, 4 fires in tropical savanna woodlands, and 2 samples from wildfires in NE Victoria.
Total emission factors observed in the field burns ranged from 0.1 to 2.9 pg TEQ (g fuel)-1 for total polychlorinated dibenzodioxins and dibenzofurans (PCDD/PCDF) and polychlorinated biphenyls ( PCBs) with means of 0.9, 1.2, 0.5 and 1.1 pg TEQ (g fuel)-1 for total PCDD/PCDF for prescribed fire, savanna fires, wildfires, and sugar cane fires, respectively. These are comparable to the minimum of the previous estimates of 0.5 to 30 pg TEQ (g fuel)-1. Dioxin was the major component (70%) with furans and PCB contributing a further 20 and 10%, respectively. The congener patterns were not uniform across all states. Emissions from cane fires and prescribed forest fires in Queensland were dominated by OCDD; in Victoria, WA and, to some extent NT, the lower chlorinated homologue groups were equally important. The main furan homologue group was TCDF and the main toxic congeners were 2,3,7,8-TCDF and 1,2,3,7,8-PeCDF. The number of observations was too small to demonstrate significant differences between classes of field burns; however, one significant and unexpected outcome was that when individual wildfires and prescribed burns were compared then prescribed burns were the stronger dioxin emitters.
Laboratory tests do not adequately simulate the combustion processes occurring in the field. The dioxin and furan emissions from the laboratory tests were substantially different to the field measurements. The laboratory burns were conducted on cereal straw, native sorghum, sugar cane trash and forest leaf litter in the test corridor at the CSIRO fire-testing laboratory. Fuel of known provenance was laid on the floor of the corridor to a loading of either 20 or 30 t ha-1. Emission rates for the grass fuels of straw, sorghum and sugar cane, were high, averaging 17, 35 and 5 pg TEQ (g fuel)-1, respectively, and ranged from 1.5 to 59 TEQ (g fuel)-1. These rates are 10-times higher than the field measurements but comparable with previously reported estimates from laboratory tests. In contrast to the field burns, the PCDD homologue groups contributed less than 30% of total mass emissions and 35 to 40% of the TEQ emissions. The most abundant homologue groups were TCDF and TCDD. Emission rates from leaf litter combustion in the laboratory ranged from 0.1 to 0.9 pg TEQ (g fuel)-1 and averaged less than a third of the field observations. Congener profiles for leaf litter burns in the laboratory were weighted toward furans.
The congener profiles observed in the laboratory tests were consistent with profiles observed in emissions from wood combustion in domestic heaters and small industrial furnaces. The field measurements particularly from SE Queensland were consistent with published field measurements from prescribed fires and soil congener profiles. The key difference between the field and laboratory emissions may be the duration for which the smoke plume remains at high temperature. In field burns, air entrained into the smoke plume rapidly cools to temperatures that will not support the heterogeneous reactions required for dioxin synthesis. In wood combustion appliances, where the combustion gases are confined within the appliance or flue, they remain at temperatures suitable for dioxin synthesis. A similar situation probably occurs during laboratory tests with grass fuel combustion, when the corridor temperatures remained above 200 ºC for the duration of the fire. However, the slower combustion rate of leaf litter produces a thermal environment intermediate between the extremes of confined grass burns and open field burns resulting in intermediate congener profiles.
Using the measured field emission rates, the total emission of PCDD/PCDF and PCBs from agricultural, forest and savanna fires in Australia were calculated and compared with the previous estimates, which were reported for the base year of 1994. The revised emission estimate for 1994 is 142 g TEQ with 95% confidence range of 31 to 494 g. This is 70% lower than the previous estimate of 72 to 1,708 g, which was based on emission factors sourced from overseas studies. Savanna fires accounted for 83% of these emissions. Wildfires, prescribed fires, and agricultural fires emit 12 g TEQ (range 6 to 23 g TEQ at 95% confidence limit) which is 95% less than previous estimate (10 to 470 g TEQ. Total emissions increased from 140 g TEQ in 1990 to 229 g TEQ in 2001, entirely due to increased savanna fire activity. This increase may be an artifact arising from a change in the methods used to estimate fire scar areas, and is currently under investigation. Emissions from wildfires, prescribed fires and agricultural fires were unchanged over this period.