Volume II Chapter 3.0 Pages 8 of 10 page next page 9

3.6.13. Motor Vehicle Fuel Combustion 3-134

3.6.14. Wood Burning at Residences 3-143

3.6.15. Industrial Wood-Burning Facilities 3-146

3.6.16. Wood Burned in Forest Fires 3-147

3.6.13. Motor Vehicle Fuel Combustion

Some of the first evidence that CDD/CDFs might be created during the combustion processes in gasoline- and diesel-fueled engines came from Ballschmiter et al. (1986) who measured these compounds in used motor oil in Germany. Incomplete combustion and the presence of a chlorine source in the form of additives in the oil or the fuel (such as dichloroethane or pentachlorophenate) were speculated to lead to the formation of CDDs and CDFs. The isomeric patterns were characterized as typical of combustion processes.

Marklund et al. (1987) provided the first direct evidence of these compounds in car emissions based on tailpipe measurements on four Swedish cars running on leaded gasoline. They found 20 to 220 pg of TEQ per kilometer driven. For this study, a nonleaded gasoline was used to which was added tetramethyl lead (0.15 grams of lead per liter or 0.57 grams per gallon) and dichloroethane (0.1 g/L as a scavenger).

The fuel used may not accurately represent commercial fuels which typically contain a mixture of chlorinated and brominated scavengers (Marklund et al., 1990). Also, the lead content of the fuel used (0.15 g lead/L), although the normal lead content for Swedish fuels (Marklund et al., 1990), is higher than the lead content of leaded gasoline in the United States during the late 1980s (lowered to 0.10 g lead/gallon or 0.026 g lead/L effective January 1, 1986).

For two cars running on unleaded gasoline, CDD/CDF emissions were below the detection limit which corresponded to approximately 13 pg of TEQ per kilometer driven. Table 3-33 presents a summary description of this study and subsequent studies discussed below.

Bingham et al. (1989) also analyzed 2,3,7,8-substituted CDD/CDFs in automobile exhausts. Four cars using leaded gasoline (0.45 g/L tetramethyllead, 0.22 g/L dichloroethane and 0.2 g/L dibromoethane) were tested and one car using lead free gasoline was tested. Only HpCDD and OCDD were detected in the exhaust from the vehicle using lead-free fuel. The total TEQ emission rate, based on these detected congeners, was 1 pg/km; the detection limit for the other 2,3,7,8-substituted CDD/CDFs was a combined 28 pg TEQ/km.

2,3,7,8-TCDF was detected in the exhaust of two of four cars using leaded fuel. OCDD was detected in the exhaust from three of the cars and PeCDF and HpCDD were each detected in the exhaust from one car. TEQ emission rates for the cars using leaded fuel, based on detected congeners, was 5 to 39 pg/km.

Haglund et al. (1988) sampled exhaust gases from three different vehicles (two cars fueled with leaded and unleaded gasoline, respectively, and a heavy diesel truck) for the presence of brominated dibenzo-p-dioxins (BDD) and brominated dibenzofurans (BDF). The authors concluded that the dibromoethane scavenger added to the tested gasoline probably acted as a halogen source. TBDF emissions measured 23 ng/km in the car with leaded gasoline and 0.24 ng/km in the car with unleaded gasoline. TBDD and PeBDF emissions measured 3.2 and 0.98 ng/km, respectively in the car with leaded gasoline. All BDD/Fs were below detection limits in the diesel truck emissions.
table Table 3-33 Descriptions and Results of Vehicle Emission Testing Studies for CDDs and CDFs
More recently, Marklund et al. (1990) tested gasoline- and diesel-fueled vehicles, measuring CDD/CDF emissions before and/or after the muffler of Swedish vehicles (including new and old vehicles).

Marklund et al. (1990) reported the emission results in units of pg TEQ/L of fuel consumed and also in units of pg TEQ/km driven during the test.

Based on the test driving cycle employed (i.e., 31.7 km/hr as a mean speed; 91.2 km/hr as a maximum speed; and 17.9 percent of time spent idling), Marklund et al. (1990) observed a fuel economy of approximately 9 to 10 km/L or 22 to 24 miles/gallon.
expand table Table V2 3-33
The following measurements were reported:

. leaded gas/cars/before muffler: 2.4 to 6.3 pg TEQ/km (or 21 to 60 pg TEQ/L of fuel consumed)
. leaded gas/cars/in tailpipe: 1.1 to 2.6 pg TEQ/km (or 10 to 23 pg TEQ/L).
. lead-free gas/catalyst-equipped car/in tailpipe: 0.36 pg TEQ/km (or 3.5 pg TEQ/L)
. lead-free gas/cars/before muffler: 0.36 to 0.39 pg TEQ/km (or 3.5 pg TEQ/L)
. diesel fuel/heavy-duty truck/before muffler: not detected (i.e., less than 100 pg TEQ/L)

Regarding the diesel fuel measurement, the authors pointed out that the test fuel was a reference fuel and may not be representative of commercial diesel fuel. Also, due to analytical problems, a much higher detection limit (about 100 pg TEQ/L) was employed in the diesel fuel tests than in the gasoline tests (5 pg TEQ/L). Further uncertainty is introduced by the fact that diesel emission samples were only collected prior to the muffler. The TEQ levels in exhaust gases from older cars using leaded gasoline were up to six times greater when measured before the muffler than after the muffler. No muffler-related difference in new cars running on leaded gasoline or in old or new cars running on unleaded gasoline were observed.

Hagenmaier et al. (1990) ran a series of tests on gasoline engines for light duty vehicles in Germany. The following average TEQ emission rates per liter of fuel consumed were found:

. Leaded fuel: 1.083 ng TEQ/L
. Unleaded fuel (catalyst-equipped): 0.007 ng TEQ/L
. Unleaded fuel (not catalyst-equipped): 0.051 ng TEQ/L
. Diesel fuel: 0.024 ng TEQ/l

Several European studies have evaluated CDD/CDF emissions from vehicles by measuring the presence of CDD/CDFs in tunnel air. This approach has the advantage that it allows random sampling of large numbers of cars, including a range of ages and maintenance levels.

The disadvantage of this approach is that it relies on indirect measurements (rather than tailpipe measurements) which may introduce unknown uncertainties and make interpretation of the findings difficult. Concerns have been raised that the tunnel monitors are detecting resuspended particulates which have accumulated over time, leading to overestimates of emissions.

Also, the driving patterns encountered in these tunnel studies are more or less steady state driving conditions rather than the transient driving cycle and cold engine starts that are typical of urban driving conditions and which may affect emission levels. Wevers et al. (1992) found that CDD/CDF levels inside a Belgium tunnel were about twice the levels in ambient air and estimated the average level in vehicle emissions as 42 to 45 pg TEQ/Nm3.

Rappe et al. (1988) conducted a similar tunnel study in Sweden and Oehme et al. (1991) conducted a similar study at a tunnel in Norway, the preliminary results of which were reported by Larssen et al. (1990). The Oehme et al. (1991) study estimated emissions for light duty and heavy duty vehicle classes. This was completed by counting the number of light duty vs. heavy duty vehicles passing through the tunnel during the study.

The mean emission rate estimates from this study are:

. Light-duty vehicles using gasoline (approximately 30 percent using leaded gas): 0.28 ng TEQ/km
. Heavy-duty diesel trucks: 5.1 ng TEQ/km

These mean values are the averages of the emission rates corresponding to two operating modes: vehicles moving uphill on a 3.5 percent incline and vehicles moving downhill on a 3.5 percent decline. The TEQ emission rates for the two modes differ by an order of magnitude for both light and heavy duty vehicles. Although Oehme et al. (1991) reported results in units of Nordic TEQs rather than I-TEQs, the results in I-TEQ should be virtually identical because the only difference between the two TEQ schemes is the factor assigned to 1,2,3,7,8-PeCDF (0.1 in Nordic and 0.05 in I-TEQ), a minor component of the toxic CDD/CDFs measured in the tunnel air.

Virtually no testing of vehicle emissions in the United States for CDD/CDFs has been published. In 1987, the California Air Resources Board (CARB) produced a draft report on the testing of the exhausts of four gasoline-powered cars and three diesel fuel-powered vehicles (one truck, one bus, and one car) (CARB, 1987).

However, CARB has indicated to EPA that the draft report should not be cited or quoted to support general conclusions about CDD/CDFs in motor vehicle exhausts because of the small sample size of the study and because the use of low rather than high resolution mass spectrometry in the study resulted in high detection limits and inadequate selectivity in the presence of interferences (Lew, 1993).

CARB did state that the results of a single sample from the heavy-duty diesel truck could be reported because congeners from most of the homologue groups were present in the sample at levels that could be detected by the analytical method and there were no identified interferences in this sample. However, it should be noted that this test was conducted under steady state conditions and at low speeds which are not indicative of normal driving patterns. The TEQ content of this one sample was 218 pg per dry standard cubic meter (dscm) of exhaust. The CARB results suggest that diesel-fueled trucks do emit CDD/CDFs (Lew, 1993).

Jones (1993) estimated CDD/CDF emissions of the major vehicle types on the basis of the studies by Larssen et al. (1990), the 1987 draft report by CARB, and Hagenmaier et al. (1990). Using data on vehicle miles travelled in the U.S. and an assumed emission rate of 5.4 ng TEQ/km based on Larssen et al. (1990), Jones (1993) estimated that about 1000 g of TEQ were emitted from diesel vehicles nationwide in 1990. Jones (1993) also suggests that human exposure to diesel emissions are exacerbated relative to stack emissions from combustion sources on the basis that diesel emissions occur at ground level and that, unlike stack emissions, may not undergo much dilution in air before human contact occurs.

In 1973, EPA required refiners to meet a 0.5 gpg (gram per gallon) standard for the average lead content of all gasoline. EPA later replaced this standard with a standard for the lead content of leaded gasoline only. Effective November 1, 1982, large refineries were required to meet a standard of 1.10 grams per leaded gallon (gplg). Certain smaller refineries were subject to a 1.90 gplg standard until July 1, 1983, at which time they would also be subject to the 1.10 gplg standard (Federal Register, 1982).

EPA further reduced the standard to 0.10 gplg effective January 1, 1986 with a[n] interim standard of 0.5 gplg effective July 1, 1985 (Federal Register, 1985). The Clean Air Act Amendments of 1990 imposed further restrictions as follows: "After December 31, 1995, it shall be unlawful for any person to sell, offer for sale, supply, offer for supply, dispense, transport, or introduce in commerce, for use as fuel in any motor vehicle any gasoline which contains lead or lead additives."

In 1985, the year before the phasedown of leaded gasoline from 1.10 gplg to 0.10 gplg, approximately 1,774 billion miles (2,855 billion km) were driven (U.S. DOC, 1992). Because leaded gasoline accounted for 35.5 percent of the gasoline supply that year (EIA, 1993) it can be estimated that 1,013 billion of these kilometers (i.e., 35.5 percent of 2,855 billion km) were driven by vehicles powered with leaded gasoline.

The U.S. Federal Highway Administration, as reported in U.S. DOC (1992), reports that 2,148 billion total vehicle miles (3,456 billion km) were driven in the U.S. during 1990. During 1990, automobiles and motorcycles accounted for 1,525 billion vehicle miles (2,454 billion km).

Trucks accounted for 617 billion vehicle miles (993 billion km) and buses accounted for 5.7 billion vehicle miles (9.2 billion km) (U.S. DOC, 1992). In 1987, diesel-fueled trucks accounted for 17.2 percent of total truck vehicle km driven; gasoline-fueled trucks accounted for the remaining 82.8 percent (U.S. DOC, 1990b).

Applying this factor (i.e., 17.2 percent) to the 1990 km truck mile estimate (i.e., 993 billion km) indicates that an estimated 171 billion km were driven by diesel-fueled trucks in 1990. It is assumed that all other vehicle km driven (3,285 billion km) were those of gasoline-powered vehicles. In 1990, leaded gasoline accounted for only 5.3 percent of total gasoline supplies (EIA, 1993). These mileage estimates are given a "high" confidence rating on the basis that they are based on U.S. Census of Transportation studies.

Using the above literature, separate emission estimates were developed for vehicles burning leaded gasoline, unleaded gasoline and diesel fuel:

. Leaded Gasoline:
In general, the literature indicates that CDD/CDF emissions occur from vehicles using leaded gasoline and that considerable variation occurs depending, at least in part, on the types of scavengers used. Marklund et al. (1987) reported emissions ranging from 20 to 220 pg TEQ/km from four cars fueled with a reference fuel (0.5 gplg) to which lead and a chlorinated scavenger were added. Marklund et al. (1990) reported much lower emissions in the tailpipe exhaust of two cars (1.1 to 2.6 pg TEQ/km) using a commercial leaded fuel (0.57 gplg).

Marklund et al. (1990) attribute the difference in the emission measurements to the different scavengers used in the two studies. Hagenmaier et al. (1990) reported TEQ emissions of 1,083 pg/L of fuel (or approximately 108 pg TEQ/km) from a car fueled with a commercial leaded fuel (lead content not reported). Bingham et al. (1989) reported emissions from four cars using gasoline with a lead content of 1.7 gplg in New Zealand to range from 5 to 39 pg/km.

The tunnel study by Oehme et al. (1991) indicated that emissions from cars could be 38 to 520 pg TEQ/km. Since most of the vehicles passing through the tunnel studied by Oehme et al. (1991) used unleaded fuels (approximately 70 percent), the emissions from leaded fuel-powered cars were possibly even higher.

On the basis of the three studies performed using commercial leaded fuel, an emission factor range of 1.1 to 108 pg TEQ/km is recommended. A "low" confidence rating is assigned to this factor range because the range is based on European fuels and emission control technologies which may differ from U.S. fuels and technology and also because the factor range is based is based on tests with only seven cars.

Combining this emission factor range with the estimates for km driven by leaded fuel-powered vehicles in 1985 (1,013 billion km or 35.5 percent of total km) and in 1990 (174 billion km or 5.3 percent of the 3,285 billion km driven by gasoline-powered vehicles) suggests that about 1.1 to 109 g TEQ/yr were emitted from vehicles using leaded fuels in 1985, the year immediately before the phasedown of leaded gasoline from 1.10 gplg to 0.10 gplg and the year when the interim standard of 0.5 gplg became effective. By comparison, the annual emission for 1990 from use of leaded gasoline is estimated to have ranged from 0.2 to 19 g TEQ.

. Unleaded Gasoline:
The literature clearly indicates that CDD/CDF emissions are much less from vehicles burning unleaded fuels. The Marklund et al. (1990) study is the only one which provided an emission factor for this class of vehicles. On the basis of this study, an emission factor of 0.36 pg TEQ/km is recommended. A "low" confidence rating is assigned to this factor because the Swedish fuels and emission control technology used in the Marklund et al. (1990) study may differ from U.S. fuels and technology and also because the emission factor is based on tests with only one catalyst-equipped car.

Combining this emission factor with the above estimates for vehicle km driven in 1990 by gasoline-powered vehicles (3,285 billion), suggests that about 1.3 g of TEQ/yr were emitted from vehicles using unleaded fuels in 1990. Based on the low confidence rating, the estimated range of potential annual emissions is assumed to vary by a factor of 10 between the low and high ends of the range. Assuming that the best estimate of annual emissions (1.3 g TEQ/yr) is the geometric mean of this range, then the range is calculated to be 0.4 to 4.1 g TEQ/yr.

. Diesel Fuel:
Very few data are available upon which to base an evaluation of the extent of dioxin emissions resulting from diesel fuel combustion. The tunnel study by Oehme et al. (1991) generated an estimated emission factor of 5.1 ng TEQ/km. A "low" confidence rating is assigned to this factor because the factor is based on Norwegian fuels and emission control technology which may differ from U.S. fuels and technology.

Also, although aggregate samples were collected representing hundreds of vehicles, the indirect method of analysis and the more or less steady state rather than transient driving conditions of the study introduce considerable uncertainty.

The results of only one tailpipe measurement (diesel fuel in a heavy-duty Swedish truck) have been published (Marklund et al., 1990) and that study reported no emissions at a detection limit of 100 pg TEQ/L. If it is assumed that the fuel economy of heavy-duty diesel vehicles is approximately 5 miles/gallon (or 2 km/L), then 100 pg TEQ/L converts to approximately 0.05 ng TEQ/km - a factor 100-fold lower than the emission rate reported by Oehme et al. (1991). Because the results of Marklund et al. (1990) are based on only one vehicle using a Swedish reference, not a commercial, diesel fuel this emission factor is also assigned a "low" confidence rating.

To obtain an estimate of the possible range of dioxin TEQ annual emissions resulting from diesel fuel use, the geometric mean of the emission factors derived from the Oehne et al. (1991) and Marklund et al. (1990) was calculated (0.5 ng TEQ/km). Combining this emission factor with the above estimate for vehicle kms driven in 1990 in the United States by diesel-fueled trucks (171 billion km) yields an annual emission estimate of 85 g TEQ/yr.

Based on the low confidence ratings, the estimated range of potential annual emissions is assumed to vary by a factor of 10 between the low and high ends of the range. Assuming that the best estimate of annual emissions (85 g TEQ/yr) is the geometric mean of this range, then the range is calculated to be 27 to 270 g TEQ/yr.

3.6.14. Wood Burning at Residences

Measurable levels of TCDDs have been found in chimney soot and bottom ash from wood-burning stoves and fireplaces (Clement et al., 1985b; Wenning et al., 1992). Chimney deposits from residential wood burning have been found to have CDD/CDF congener profiles similar to those in flue gases from municipal waste incineration (Bacher et al., 1992). Bacher et al. (1992) found concentrations of 2,3,7,8-substituted CDF and CDD congeners in soot from wood burning ranging from 40 to 930 ng/kg and from 30 to 150 ng/kg, respectively. Bacher et al. (1992) reported that the CDFs dominated the CDDs by a factor of 5 to 10 and that the lower chlorinated CDDs/CDFs (mono-through tri-) dominated the more chlorinated CDDs/CDFs. The TEQ content of the chimney soot was 720 ng/kg (Bacher et al., 1992).

Nestrick and Lamparski (1983) conducted a study of CDD formation in residential wood-burning chimneys in different areas of the United States. The results of their survey are presented in Table 3-34. As seen in Table 3-28, the eastern U.S. had overall higher estimated levels of TCDD generation than did the central or western United States. Levels of TCDDs in chimney soot ranged from 22 to 410 ng/kg for the eastern U.S., 21 to 294 ng/kg in the central U.S., and 2.9 to 28 ng/kg in the western U.S. Red oak and oak were the predominant fuels used in the eastern and western U.S., and ash, birch, and oak were the predominant fuels used in the central U.S.

Two studies have recently become available which provide direct measurement of CDD/CDF in emissions from wood stoves. These studies are summarized below.

Schatowitz et al. (1993) measured CDD/CDF in the emissions of a variety of residential wood burners in Switzerland. The study included three types of burners (household stoves, automatic chip burners, and wood boilers) and a variety of wood fuels (natural beech wood, natural wood chips, chipboard chips, waste wood chips, charcoal and household waste). The following emission factors were derived:

• household stove with open door burning natural beech wood: 0.77 ng TEQ/kg

• household stove with closed door burning natural beech wood: 1.25 ng TEQ/kg

The open door stove can be assumed to be representative of fireplaces since both have an uncontrolled draft. Also, Schatowitz et al. (1993) measured emissions from wood burning fireplaces and report the same flue gas concentration as found with the open door wood stove (i.e., 0.064 ng TEQ/m3). All of the toxic congeners of CDD/CDF were found at levels above the detection limit.

Vikelsoe et al. (1993) studied emissions of CDD/CDFs from residential wood stoves in Denmark. The wood fuels used in the experiments were seasoned birch, beech and spruce harvested in Denmark. Four different types of stoves were evaluated under normal and optimal operating conditions. ...

table Table 3-34 Average Concentrations (ppt) of TCDDs in Chimney Soot from Residential Wood-Burning Stoves in the U.S.
... Widely varying emissions were found for different fuel/stove combinations. The emissions from spruce were about twice as high as the emissions from birch and beech.

In many of the experiments the CDD/CDF emissions were higher when the stove was operated under normal conditions vs optimal conditions (minimum CO).

The weighted average (considering wood and stove types) emission factor for wood stoves in Denmark was estimated to be 1.9 ng Nordic-TEQ/kg. Based on the above studies, 1 ng TEQ/kg appears to be a reasonable average emission factor for residential wood burning.
A "medium" confidence rating was assigned to this estimate on the basis that:
expand table Table V2 3-34

(1) it is derived from only two studies;
(2) both studies used direct measurement; and
(3) although the studies were conducted in Europe, residential wood burning practices are probably sufficiently similar to apply to the United States.

In 1990, wood provided about 2.8 percent of the total primary energy consumed in the United States (EIA, 1991). Total wood energy consumption during 1990 is estimated at 2,359 trillion BTU. Assuming that 1 kg of oven-dried wood (i.e., 2.15 kg of green wood) provides approximately 19,000 BTU, then an estimated 124.2 million metric tons of oven-dried wood equivalents were burned for energy purposes in 1990 (EIA, 1991). Residential wood consumption in 1990 was estimated at 786 trillion BTU (41.4 million metric tons), or 33 percent of total U.S. consumption. Industrial fuel wood consumption in 1990 totaled 1,562 trillion BTU (82.2 million metric tons), or 66 percent of total U.S. consumption, with the majority (1,232 trillion BTU) of this fuel wood being consumed by the Paper and Allied Products Industry. 1990 consumption of fuel wood by the utility sector was approximately 11.9 trillion BTU (0.6 million metric tons) (EIA, 1991). These production estimates are given a "high" confidence rating since they are based on a detailed published study specific to the United States.

Combining the best estimate of the emission factor (1 ng TEQ/kg wood) with the mass of wood consumed annually by residences, indicates that the annual TEQ emissions from this source are about 40 grams. Based on the "medium" confidence rating assigned to the emission factor, the estimated range of potential annual emissions is assumed to vary by a factor of 5 between the low and high ends of the range. Assuming that the best estimate of annual emissions (40 g TEQ/yr) is the geometric mean of this range, then the range is calculated to be 13 to 63 g TEQ/yr.

3.6.15. Industrial Wood-Burning Facilities

Emissions of dioxin-like compounds have been measured in stack emissions from an industrial wood-burning furnace by EPA (U.S. EPA, 1987). The tested facility was located at a lumber products plant that manufactures overlay panels and other lumber wood products. The wood-fired boiler tested was a three-cell dutch oven equipped with a waste heat boiler.

During normal operation, the furnace is 100 percent fired with scrap wood from the lumber plant. The feed wood is a mixture of bark, hogged wood, and green and dry planar shavings. The composition of the feed was estimated to be wood from fir and hemlock. Nearly all the wood fed to the lumber plant had been stored in sea water adjacent to the facility, and therefore had a significant concentration of inorganic chloride.

The scrap wood fed to the boiler had not been treated with chemical preservatives, e.g., pentachlorophenol. The wood was fed to the boiler by a screw conveyor that dumps the feed into a pile in the primary combustion chamber. The furnace is operated at air in 50 percent excess of stoichiometric requirements. Boilers capture the heat of combustion and transfer the heat into steam for co-generation of energy at the plant.

The exhausted gases from the boiler pass through a cyclone and fabric filter prior to discharge from the stack. From this study, an average emission factor for CDD/CDF of 1.02 µg/kg of wood burned (range: 5.52E-01 to 1.41E+00 µg/kg), and an average emission factor for TEQ of 1.71E-02 µg/kg of wood burned (range: 7.34E-03 to 2.28E-02 µg/kg) are estimated. Emissions testing at this facility demonstrated that the fabric filter was reducing dioxin emissions by about 90 percent (U.S. EPA, 1987).

In a second study, CDD/CDF was measured in the emissions from a quad-cell wood-fired boiler used to generate electricity (CARB, 1990b). The fuel consisted of coarse wood waste and sawdust from nonindustrial logging operations. The exhaust gas passed through a multicyclone before entering the stack. This study suggests an emission factor of 5.4E-02 m g/kg for CDD/CDF. If the same TEQ to total CDD/CDF ratio is assumed as in the first industrial burner study, then an emission factor of 9E-04 m g TEQ/kg can be estimated.

To obtain an estimate of the possible range of dioxin TEQ annual emissions resulting from industrial wood-burning facilities, the geometric mean of the emission factors derived from the U.S. EPA (1987) and CARB (1990b) studies was calculated (3.9 ng TEQ/kg wood). Because test data are available for only two facilities and because the emission factors measured at these two facilities vary greatly, this emission factor was given a "low" confidence rating.

In 1990, it was estimated that 82.2 million metric tons of wood were burned in industrial furnaces ("high" confidence rating, see discussion in Section 3.6.14). Applying the above emission factor to the estimated annual mass of wood burned by industrial facilities gives an estimated TEQ emission of 320 g TEQ/yr. Based on the low confidence rating given to the emission factor, the estimated range of potential annual emissions is assumed to vary by a factor of 10 between the low and high ends of the range. Assuming that the best estimate of annual emissions (320 g TEQ/yr) is the geometric mean of this range, then the range is calculated to be 100 to 1,000 g TEQ/yr.

3.6.16. Wood Burned in Forest Fires

Based on the findings of Wenning et al. (1992), Bacher et al. (1992), and Nestrick and Lamparski (1983), indicating generation of CDDs/CDFs in ash and soot during residential wood burning, it is reasonable to hypothesize that wood burned in forest fires may also be a source of CDDs/CDFs. Support for this hypothesis is provided by Bumb et al. (1980) who, in their study on trace chemistry of fire, have shown that combustion of hydrocarbons in the presence of chlorine compounds (which are naturally found at low levels in wood) can generate CDDs and CDFs in small amounts. Also, the pre-industrial existence of CDDs and CDFs, presumably due to combustion sources, has been demonstrated in analyses of ancient human tissues and ancient aquatic sediment deposits (ECETOC, 1992).

Only one study could be found that made direct measurements of CDD/CDFs in the actual emissions from forest fires. This study by Tashiro et al. (1990) detected levels ranging from about 15 to 400 pg/m3 for total CDD/CDFs. The samples were collected from fixed collectors 10 m above the ground and from aircraft flying through the smoke. Background samples collected before and after the tests indicated negligible levels in the atmosphere.

These results were presented in the form of a preliminary report and no firm conclusions were drawn about whether forest fires are a CDD/CDF source. Coauthor Dr. Ray Clement presented the final report on this study at Dioxin '91. Clement and Tashiro (1991) reported total CDD/CDF levels in the smoke of about 20 pg/m3. The authors concluded that CDD/CDFs are emitted during forest fires but recognized that some portion of these emissions could represent resuspension from residues deposited on leaves rather than newly formed CDD/CDFs.

The concentrations presented by Clement and Tashiro (1991) cannot accurately be converted to an emission factor since the corresponding rates of combustion gas production and wood consumption are not known. As a result, three alternative approaches were considered to develop these emission factors:

Soot-Based Approach:
This approach assumes that the level of CDD/CDFs in chimney soot are representative of the CDD/CDFs in emissions, and estimates the CDD/CDF emission rate as the product of the soot level and the total particulate emission rate. This involves first assuming that the CDD/CDF levels measured by Bacher et al. (1992) in chimney soot (720 ng TEQ/kg) are representative of the CDD/CDF concentrations of particles emitted during forest fires. Second, the total particulate generation rate must be estimated. Ward et al.(1976) estimated the national average particulate emission factor for wildfires as 150 lb/ton biomass dry weight based primarily on data for head fires.

Ward et al. (1993) estimated the national average particulate emission factor for prescribed burning as 50 lb/ton biomass dry weight. Combining the total particulate generation rates with the CDD/CDF levels in soot yields emission factor estimates of 54 µg of TEQ and 18 µg of TEQ/metric ton of biomass burned in wildfires and prescribed burning, respectively. This corresponds to a range of 54 to 18 ng TEQ/kg of biomass. This estimate is likely to be an overestimate since the levels of CDD/CDF measured in chimney soot by Bacher et al. (1992) may represent accumulation/enrichment of CDD/CDFs measured in chimney soot over time, leading to much higher levels than what is actually on emitted particles.

Carbon Monoxide (CO) Approach:
CO is a general indicator of the efficiency of combustion and the emission rate of many emission products can be correlated to the CO emission rate. The Schatowitz et al. (1993) data for emissions during natural wood burning in open stoves suggests an emission rate of 10 ug TEQ/kg of CO. Combining this factor with the CO production rate during forest fires (roughly 0.1 kg CO/kg of biomass - Ward et al. (1993)) yields an emission factor of 1000 ng TEQ/kg biomass. This factor appears unreasonably high since it is even higher than the soot-based factor discussed above. Although the formation kinetics of CDD/CDF during combustion are not well understood, it appears that CDD/CDF emissions do not correlate well with CO emissions.

Wood Stove Approach:
This approach assumes that the emission factor for residential wood burning (using natural wood and open door, i.e., uncontrolled draft) applies to forest fires. As discussed in Section 3.6.14, this approach suggests an emission factor of about 1 ng TEQ/kg. This value appears more reasonable than the factors suggested by the soot and CO approaches. However, forest fire conditions differ significantly from combustion conditions in wood stoves. For example, forest fire combustion does not occur in an enclosed chamber and the biomass consumed in forest fires is usually green and includes underbrush, leaves and grass.

Given these differences and the uncertainties about the formation kinetics of CDD/CDF during combustion, it is difficult to determine whether CDD/CDF emissions would be higher or lower from forest fires than from wood stoves. Thus, although an emission factor of 1 ng TEQ/kg appears to be the best estimate that can be made currently, it must be considered highly uncertain and a "low" confidence rating was assigned to this estimate.

According to the Council on Environmental Quality's 21st Annual Report (CEQ, 1990), an average of 5.1 million acres of biomass have been burned in wildfires every year from 1950 to 1990. This value also corresponds well to the data provided by the USDA Forest Service for 1975 in which 4.4 million acres of biomass were burned in wildfires (Ward et al., 1976). Yearly estimates cited in the CEQ report (CEQ, 1990) ranged from a high of 15.5 million acres burned to an annual low of 1.8 million acres burned over the forty year time period. Additionally, 5.1 million acres of biomass were burned in 1989 during prescribed burns (Ward et al., 1993).

Prescribed burning is also known as managed or controlled burning and is used as a forest management tool under exacting weather and fuel conditions. These acreage estimates can be combined with biomass consumption rates of 10.4 tons/acre in areas consumed by wildfires (Ward et al., 1976) and 8.2 tons/acre in areas consumed in prescribed burns (Ward et al., 1993). This combination suggests a total of 53 million tons (or 48 million metric tons) of biomass are consumed annually in wildfires while a total of 42 million tons (or 38 million metric tons) of biomass are consumed annually in prescribed burns. These production estimates were assigned a "medium" confidence rating since they are based on a combination of estimates involving detailed historical data specific to the United States on acres burned but less accurate estimates of biomass burned/acre.

Combining the emission factor developed using the "wood stove" approach (i.e., 1 ng TEQ/kg biomass) with the amount of biomass consumed annually in wildfires and prescribed fires (total of 86 million metric tons) indicates that the best estimate of annual TEQ emissions from this source is 86 g. Based on the low confidence rating given to the emission factor, the estimated range of potential annual emissions is assumed to vary by a factor of 10 between the low and high ends of the range. Assuming that the best estimate of annual emissions (86 g TEQ/yr) is the geometric mean of this range, then the range is calculated to be 27 to 270 g TEQ/yr.

Whether releases from this source result in significant human exposure is questionable. If wood burning today is a major source of human exposure to CDD/CDFs, then the tissues of ancient humans (who relied on wood as a fuel source more so than do humans in industrialized settings today) should have CDD/CDF levels that are a substantial fraction of the levels found in humans today. The 1987 NHATS study indicates that the U.S. average CDD/CDF adipose tissue concentration is currently about 1000 ppt (U.S. EPA 1990d) and Schecter (1991) reports that the total CDD/CDF concentration in liver tissues today is about 400 ppt.

However, with the exception of OCDD in one sample, Ligon et al. (1989) could not detect CDD/CDFs that exceeded the analytical background (detection limit = 0.3 to 5 ppt) in the mummified muscle tissues of nine 2,800 year old Chilean Indians. Similarly, Schecter (1991) examined the livers of two frozen 400 year old Eskimo women and found only HpCDD, OCDD, and HpCDF at levels only 15 percent of current levels.