Annex 46 Complementary information to source category 6b Waste Burning and Accidental Fires

Overview of recent revisions

Similarly to the section on biomass, since the 2nd edition of the Toolkit (2005) new information has become available for the open burning of waste from several projects that have been implemented to provide better emission factors and especially taking into consideration circumstances in developing countries. The projects included field experiments in Mexico and China that added a total of 30 results to improve the emission factors for dump fires (class 1) and burning of loosely arranged domestic waste (class 3) (Solorzano-Ochoa et al. 2012, Zhang et al. 2011, UNEP 2011a). Besides providing new emission factors, the characterization of the activity has also been made more specific and a protocol and a sampler for undertaking field sampling have been developed. Although the emission factors for PCDD/PCDF and dioxin-like PCB scatter across a large range between measurements, the confidence in these data is relatively high.

Table III.46.1 Dioxin-like PCB emission factors for source category 6b Open Burning of Waste and Accidental Fires

6b Open Burning of Waste and Accidental Fires Emission Factors (µg TEQ/t material burned)
  Classification Air Water Land Product Residue
1 Fires at waste dumps (compacted, wet, high organic carbon content) 30        
2 Accidental fires in houses, factories          
3 Open burning of domestic waste 2        
4 Accidental fires in vehicles (µg TEQ per vehicle)          
5 Open burning of wood (construction/ demolition)          

Although first, indicative measurements were undertaken for hexachlorobenzene and pentachlorobenzene, these emission factors are not proposed for inclusion into this Toolkit since the sampling resin (polyurethane foam) has not been adapted to these more volatile POPs and breakthrough or losses of analytes cannot be excluded. A literature study identified new information for class 4, accidental fires of cars that confirmed the present EFAir. For order-of-magnitude assessments, the earlier EFAir was changed to 100 µg TEQ per vehicle burned.

Derivation of emission factors

Release to Air

Open burning of mixed municipal solid waste is addressed in two classes within this sources category: 6b1 - Fires at waste dumps (compacted, wet, high organic carbon content) and 6b3 – Open burning of domestic waste. Both waste content and open burning of waste are highly variable and difficult to characterize. Emission factors for open burning of waste from the early Toolkit versions (1st and 2nd edition of the UNEP Toolkit) were based on experiments from laboratory simulations (USEPA 1997b, Gullett et al. 1999, Lemieux et al. 2003).

For open burning of domestic waste, Zhang et al. (2011) measured PCDD/PCDF releases to air from 20 field burns of urban, semi-urban and rural wastes in China and Mexico using a newly developed sampling system. In China, the contents of each type of waste were based on literature surveys and field observation of wastes. For the ten field burns in China, an average EFAir of 18 ng TEQ/kg (range 3.0 to 51 ng TEQ/kg) was determined. In Mexico, waste compositions were based on results of sampling and analyzing materials at landfills in rural, semi-urban and urban-industrial areas of Mexico. For the ten field burns in Mexico, an average EFAir of 20.4 ng TEQ/kg (range 35 to 650 ng TEQ/kg) was derived.

In Sweden, Hedman et al. (2005) burned garden waste and refuse derived fuel (“municipal waste where the combustible fractions (e.g. paper, textile and soft plastics) had been mechanically sorted out from noncombustible waste and decomposable material at a waste sorting plant”) in open steel barrels and on a steel plate. They reported air emission factors of 4-72 ng TEQ/kg, with a median value of 20 ng WHO-TEQ/kg . More specifically, Hedman et al. (2005) reported EFAir of 16-18 ng TEQ/kg.

For fires at landfills and waste dumps, a number of simulations and small scale studies have investigated varying waste composition, in particular chlorine and metal content to explain the variability in emission factors for PCDD/PCDF (e.g., Ikeguchi and Tanaka 2000, Nakao et al. 2006, Gullett et al. 1999). They have not generated conclusive results and recent experiments at laboratory scale in burn huts showed that small scale experiments do not necessarily mimic field situations (UNEP 2010b). Therefore, for the determination of emission factors for open burning of waste at landfills and waste dumps in this Toolkit, only large scale, field experiments were further assessed.

Thirty recent field experiments were conducted in four campaigns (Solorzano Ochoa et al. 2011, Zhang et al. 2011), three in Mexico and one in China. EFAir for PCDD/PCDF (in TEQ) spanned over three orders of magnitude, i.e., ranged from 13 µg TEQ/t Cburned to 14,000 µg TEQ/t Cburned. In one of the Mexican campaigns, although only six experiments were performed, extremes of conditions (stirred combustion to compacted, moistened, waste without stirring) were performed on the same waste composition. In these experiments, the EFAir spanned over two orders of magnitude (from 290 µg TEQ/t Cburned to 14,000 µg TEQ/t Cburned). The stirred experiments generated the lowest emission factors and the unstirred, compacted experiment with added water yielded the highest (Solorzano Ochoa et al. 2011, Zhang et al. 2011). Higher EFAir are observed at higher ratios of CO/CO2; both occurred at later stages of the burning experiments when mass loss was lowest.

These experiments suggest that unattended, smoldering, poorly oxygenated open burning yields the highest emissions. Better combustion conditions, induced by stirring (presumably better oxygen supply and higher combustion temperatures), give rise to less-polluting open burning. Such conclusion is consistent with the recommendations that were provided in the BAT&BEP guidance document elaborated for the Stockholm Convention (UNEP 2007). This also means that only small piles could be stirred and large-scale open burnings of entire dumps may be analogous to the unstirred experiments.

The results from the Mexican and Chinese field campaigns support the use of two distinct classes within the Toolkit source category 6b: compacted repositories vs. loosely arranged piles. Promoting good combustion conditions seems to be a solid and basic principle; however, considerable additional work should be done to fully understand the mechanistic factors that drive the emission factors towards the higher or lower end of the scale. These experiments also allowed derivation of EFAir for dl-PCB whereas the sampling method was not optimized to securely derive EFAir for the more volatile unintentional POPs, namely hexachlorobenzene (HCB) and pentachlorobenzene (PeCBz).

Since the last version of the Toolkit, no additional information was found for class 6b2 – Accidental fires in houses, factories. Materials burned and conditions experienced in accidental fires vary enormously. Limited data are available giving air emissions from such fires. Given the difficulties of measuring air emissions, most studies have used laboratory simulations or measured PCDD/PCDF in soot and residues. Some studies have considered air emissions from the burning of PVC only and provided air emission factors based on soot production in the range 40-3,500 μg TEQ/t (based on summary by Carroll 1996). These factors were based on results from laboratory and real fire samples and assume that 20% of the PVC remains unburned and that all emissions to air are captured in the soot. Merk et al. (1995) burned wood and PVC in a closed room and measured levels of PCDD/PCDF in the air/flue gas in the room as well as in wipe samples from the walls. Assuming all the air in the room was contaminated at the levels measured an EFAir of 560 μg TEQ/t of PVC or 51 μg TEQ/t of the wood/PVC mixture is obtained. Further it was assumed that deposited soot is additional to this a further 2,200 μg TEQ/t PVC or 200 μg TEQ/t PVC/wood mixture was produced. Ikeguchi and Tanaka (1999) provide air emission factors for the open burning of various wastes (220 μg TEQ/t for scrap tires, 1,000 μg TEQ/t for electric wire tube and 6,600 μg TEQ/t for agricultural plastics (PVC)). Testing in Denmark of PCDD/PCDF released from burning chemicals (Vikelsoe and Johansen 2000) showed an enormous range of air emission factors (at 500°C) from 1 μg TEQ/t for dichlobenil, up to 740,000 μg TEQ/t for pentachlorophenol (PCP), and 100 μg TEQ/t for PVC. At 900°C yields were greatly reduced. After a review of accidental fire data, air emission factors for residential fires of 83 μg TEQ/t to air, 83 μg TEQ/t in locally deposited soot and for industrial fires 500 μg TEQ/t to air and 500 μg TEQ/t in locally deposited soot were developed for Germany (LUA 1997).

Given the wide range of materials considered under accidental fires and the wide range of possible emission factors an initial estimate can be made by applying an EFAir of 400 μg TEQ/t to accidental fires.

Some new information was found for class 6b4 – Accidental fires in vehicles. Two studies were taken into account when deriving the emission factor for the release of PCDD/PCDF from fires in vehicles. Wichmann et al. (1995) carried out experiments in a tunnel with an old car (1974), a new car (1988), a subway car and a railway carriage. Emissions to air were estimated from PCDD/PCDF deposited inside the tunnel and no estimate was made of PCDD/PCDF, which may not have been deposited. Emissions were 32 μg TEQ for the old car, 44 TEQ μg for the new car, 2,000 μg TEQ for the subway car and 9,200 μg TEQ for the railway wagon. For the purposes of making an initial estimate a composite emission factor is developed assuming 49.5% of vehicle fires involve “old” cars, 49.5% involve new cars and 0.5% each of vehicles represented by subway cars and railway wagons giving a factor of 94 μg TEQ per incident. Lönnermark and Blomqvist (2006) measured emissions from an automobile fire. Three separate full scale fire tests have been undertaken: a fire ignited and developed in the engine compartment; a fire ignited inside the coupé, that was extinguished in the early stages; and a similar fire ignited inside the coupe´ that was allowed to spread until the entire vehicle was involved in the fire. The emissions to air were from 71.1 g-I-TEQ per car to 86.8 g I-TEQ per vehicle. Combining the results from the Wichmann et al. (1995) and Lönnermark and Berqvist (2006), an EFAir of 100 g per vehicle was assigned. This EFAir is per incident not per ton of material.

No change was undertaken for class 6b5. It is relatively common to see fires used to dispose of wastes from construction and demolition – these are often predominantly burning wood but may also be used to dispose of other materials. Again little information is available either on the amounts burned or on emissions. When Wasson et al. (2005) burned aged and weathered chromated copper arsenate (CCA) treated wood was burned in USEPA’s burn hut, they reported PCDD/PCDF levels averaging 1.7 ng TEQ/kg of treated wood burned. Studies in Japan can be used to make an initial estimate of a suitable emissions factor to be applied – 60 μg TEQ/t (Ikeguchi and Tanaka 1999).

Release to Water

There is little information available on the release of PCDD/PCDF to water from these types of fires. The releases will vary considerably but highest releases may result from the use of water to extinguish a fire or rain falling on a fire site and washing material into water courses. Since these processes will depend on local circumstances it is not possible to provide emissions factors but the issue may be important.

Release to Land

Based on the recent experiments, an EFLand (or EFResidue) as derived from the experiments by Lemieux et al. (2003) were revised as follows: for class 1, an EFLand of 10 μg TEQ/t waste and for class 3 an EFLand of 1 μg TEQ/t waste are proposed.

The EFLand for classes 2, 4, and 5 remain unchanged. For accidental fires there is little information on levels of PCDD/PCDF in residues. A wide range of concentrations has been measured but there is often insufficient information to estimate an emission factor since the amounts of ash produced are not known. In Germany, an estimate was made that gave emission factors in residues (including deposited soot) of 1,000 μg TEQ/t for industrial fires and 350 μg TEQ/t for residential fires (LUA 1997). As an approximation and to make an initial estimate, an emission factor of 400 μg TEQ/t is used giving equal PCDD/PCDF in air emissions and in residues on average from the fires considered. For vehicle fires the limited testing in Germany (Wichmann et al. 1995) gave amounts of PCDD/PCDF left in residues, a composite emission factor is used to make an initial estimate – 18 μg TEQ per incident (using the same assumptions as above). Note, this emission factor is per incident not per unit mass. For fires involving construction and demolition wood, no emission factors were found. To make a preliminary estimate, an emission factor of 10 μg TEQ/t wood burned is suggested (from UK work on industrial wood combustion, Dyke et al. 1997). Note that treated wood, mixed fire loads and poor conditions may increase the amount of PCDD/PCDF in residues considerably.

Release in Products

No EFProduct is provided. No product is expected.

Release in Residues

No EFResidue is provided. Residues are assumed to be releases to land since they are typically left in place.

Detailed Methodology

Emission factors to air were calculated as ng TEQ per kg of carbon burned (ng TEQ/kg Cburned). Cburned is calculated from the carbon content of CO and CO2 resulting from combustion and are assumed to comprise the totality of airborne carbon, with other forms negligible by comparison. Emission of PCDD/PCDF per ton of waste (μg TEQ/t waste) is calculated by multiplying EFAir (ng TEQ/kg Cburned) by the carbon content of the waste and the experimental carbon oxidation factor (COF) (Fiedler et al. 2009).

A new practical approach has been developed to allow inventory developers to better characterize the activity, i.e., estimate the mass of waste, which is burned in the open air. The new method takes the whole amount of waste present for the burn event into account. The new method is based on the fact that not all organic carbon that is present in the original waste will be burned; of the carbon burned, the majority is converted to carbon dioxide and carbon monoxide; a much smaller portion to PCDD/PCDF and other organic hydrocarbons. From experimental data, it was concluded that approximately only 40% by weight of the waste is combustible carbon. In line with global climate calculations of biomass combustion, a carbon oxidation factor of 58% was applied, meaning that only 58% of the 40% combustible carbon burns, resulting in an estimated 23% of the original carbon weight is actually combusted. When the carbon content in the waste is higher and better burn-out is obtained, the overall COF resulted in 42%. Applying the lower and the higher COF to the experimental set in Mexico, the EFAir would be obtained (Table II.46.2).

Table III.46.2 Derivation of EFAir using two different COF

EFAir (ng TEQ/kg Cburned) Conversion factor Reference
23% 42%
823 189 346 Gullett et al. 2010
14,000 3,220 5,880 Solorzano Ochoa et al. 2011
660 152 277 Soloarzano Ochoa et al. 2011
290 67 122 Solorzano Ochoa et al. 2011
870 200 365 Solorzano Ochoa et al. 2011
950 219 399 Solorzano Ochoa et al. 2011
950 219 399 Solorzano Ochoa et al. 2011

An initial assessment of national release inventories made with the UNEP Toolkit has shown that open burning of biomass, such as forest, bush and grassland fires, burns in agriculture, and of waste are major sources of PCDD/PCDF in developing countries. Among the ten source groups, Source Group 6 Open burning contributes with an average of 61% to the total TEQ of all emissions to air; a second contributor is Source Group 1, mainly due to incineration of medical waste with approximately 12% and a third is Source Group 3 energy conversion and heating/cooking with approximately 7%. When PCDD/PCDF release inventories are compared, a hierarchical cluster analysis shows that Source Group 6, Open Burning, makes the difference between all inventories (Fiedler 2011).

Figure III.46.1 Dendrogram of the ten source groups in the UNEP Toolkit applied to 60 PCDD/PCDF inventories demonstrating similarities and dissimilarities

A total of 41 countries reported their releases of PCDD/PCDF – in g TEQ per year - to Air and Land for the open burning of biomass and waste (Table III.46.3, Fiedler 2011). The total releases were 18,363 g TEQ and reference year, whereby 8,958 g TEQ were emitted to air and 9,405 g TEQ to land. Considering the contribution from source categories 6a and 6b, 4,610 g TEQ were from biomass burnings and 13,753 g TEQ from waste burnings or accidents.

Noteworthy is that nine countries reported zero for releases from biomass (Guatemala, Pakistan, Tajikistan) or waste burnings (Bolivia, Cameroon, China, Honduras, Montenegro, Ukraine, Venezuela), respectively. Besides the difficulties to quantify the amount of material consumed in annual fires or doubts on the applicability of the emission factors, legal implications may be the driver for setting annual emissions to zero.

Table III.46.3 Summary of descriptive statistics for the assessment of annual releases of PCDD/PCDF (expressed as g TEQ/yr) from open burning of biomass and waste

Fuel Biomass Waste
Release Vector Air Land Air Land
Mean 24% 16% 32% 29%
Median 9% 5% 32% 34%
Std Dev. 29% 23% 28% 26%
Variance 8% 5% 8% 7%
25th Percentile 1% 0% 1% 0%
75th Percentile 56% 28% 48% 52%