National Dioxin Study of Combustion Sources conducted in 1986 (EPA, 1987d). The tested facility was judged by EPA to be typical of the industry. These plants operate a furnace to prepare used steel 55-gallon drums for cleaning to base metal. The cleaned drums are repaired, repainted, relined and sold for reuse. The used drums processed at the tested facility were from the petroleum and chemical industry. The drum burning process subjected the used drums to an elevated temperature in a tunnel furnace for a sufficient time so that the paint, interior linings, and previous contents were burned or disintegrated. The furnace was fired by auxiliary fuel. Used drums were loaded onto a conveyor that moved at a fixed feed rate. As the drums passed through the preheat and ignition zone of the furnace, additional contents of the drums drained into the furnace ash trough. A drag conveyor moved these sludges and ashes to a collection pit. The drums were air cooled as they exited the furnace. Exhaust gases from the burning furnace were drawn through a breaching fan to a high-temperature afterburner. The homologue profile for drum and barrel reclamation furnaces, shown in Figure 3-1, was developed from EPA stack tests of this operation (EPA, 1987d).

4. Medical Waste Incinerators: The State of California Air Resources Board (CARB) has stack tested a number of hospital waste incinerators in southern California (CARB, 1990). Congener-specific emissions of PCDD/Fs were measured in the stack gas emissions of 7 facilities. Figure 3-1 displays the homologue profile constructed from the average emission of three facilities tested by CARB identified as facilities A - C in their summary overview of emissions (CARB, 1990).

5. Scrap Electric Wire Incineration: Dioxin-like compounds emitted to the air from scrap electric wire incineration were measured from a facility during EPA's National Dioxin Study of combustion sources (EPA, 1987d). The objective of wire incineration is to remove the insulating material and reclaim the metal (e.g., copper, silver, and gold) comprising the electric wire, hence these facilities are sometimes referred to as wire reclamation incinerators. The reclaimed metal is then sold to a secondary metal smelter. The tested facility was judged by EPA to be typical of this industry. Insulated wire and other metal-bearing scrap material were fed to a combustion unit where incineration of the material was assisted by the combustion of natural gas. The estimated temperature during combustion was 650° C, and combustion transpired in both a primary and secondary chamber. The tested facility was equipped with a high temperature afterburner to further destroy organic compounds entrained in the combustion gases prior to discharge to the air from the stack. Figure 3-1 displays the homologue distribution developed from this single facility.

6. Automobile Tire Incineration: Homologue emissions factors shown in Figure

3-1 were developed from an automobile tire incinerator stack tested by the State of California Air Resources Board (CARB, 1991). The facility consists of two excess air furnaces equipped with steam boilers to recovery the energy from the heat of combustion. Discarded whole tires are fed to the incineration units at a rate of 3000 kg/hr. The furnaces are equipped to burn natural gas as auxiliary fuel. The steam produced from the boilers is used to drive electrical turbine generators to produce 14.4 megawatts of electricity. The facility is equipped with a dry acid gas scrubber and fabric filter for the control of emissions prior to exiting the stack. These devices are capable of greater than 95% reduction and control of dioxin-like compounds prior to discharge from the stack.

7. Industrial Wood-Burning Facilities: The homologue profile shown in Figure 3-1 for this source category was developed from measurements of stack emissions from an industrial wood-burning furnace (EPA, 1987d). The tested facility was judged by EPA as being typical of these combustion technologies. The 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% fired with scrap wood from the lumber plant. The feed wood is typically a mixture of bar, hogged wood, and green and dry planar shavings. For the stack test from which the homologue profile was developed, the feed was mostly 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, such as 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 was operated at air in 50% excess of stoichiometric requirements. Boilers captured the heat of combustion and transfered the heat into steam for co-generation of energy at the plant. The exhaust gases from the boiler passed through a cyclone and fabric filter prior to discharge from the stack. The facility was equipped with a cyclone and fabric filter to control emissions. Emissions testing at this facility demonstrated that the fabric filter was reducing dioxin emissions by about 90% (EPA, 1987d).

8. Metal Reclamation Plants: Metal reclamation plants are secondary metal smelting facilities which include secondary copper smelters, secondary aluminum smelters, secondary magnesium smelters, and secondary ferrous smelters. The only complete information with regard to the potential stack emission of dioxin-like compounds is from a secondary copper smelter tested by EPA during the National Dioxin Study (EPA, 1987d). The homologue profile shown in Figure 3-1 was developed from this facility. The tested facility was a secondary copper smelter that recovers copper and precious metals from copper and iron-bearing scrap. The copper and iron-bearing scrap was fed to a blast furnace, which produced a mixture of slag and black copper. The blast furnace was a batch-fed cupola furnace. Four to five tons of metal-bearing scrap were fed to the furnace per charge, with materials typically being charged 10 to 12 times per hour. Coke was used to fuel the furnace, which represented 14% (by wt) of the total feed. During the dioxin stack tests, the feed consisted of electronic telephone scrap and other plastic scrap, brass and copper shot, iron-bearing copper scrap, precious metals, copper bearing residues, refinery by-products, converter furnace slag, anode furnace slag, and metallic floor cleaning material. Oxygen enriched combustion air for combustion of the coke was blown up through the bottom of the furnace. At the top of the blast furnace were four natural gas-fired afterburners to aid in completing combustion of the exhaust gases. Particulate emissions were controlled by fabric filters, and the flue gas then was discharged into a common stack.

9. Kraft Black Liquor Recovery Boilers: EPA stack tested three kraft black liquor recovery boilers for the emission of dioxin in conjunction with the National Dioxin Study (EPA, 1987d). The three sites were judged by EPA to be typical of Kraft black liquor recovery boilers, and the homologue profile shown in Figure 3-1 was derived from these three sites. These sources are associated with the production of pulp in the making of paper using the Kraft process. In this process, wood chips are cooked in large vertical vessels called digesters at elevated temperatures and pressures in an aqueous solution of sodium hydroxide and sodium sulfide (Someshwar and Pinkerton, 1992). Wood is broken down into two phases: a soluble phase containing primarily lignin, and an insoluble phase containing the pulp. The spent liquor (called black liquor) from the digester contains sodium sulfate and sodium sulfide that the industry finds beneficial in recovering for reuse in the Kraft process. In the recovery of black liquor chemicals, weak black liquor is first concentrated in multiple-effect evaporators to about 65% solids. The concentrated black liquor also contains 0.5% - 4% weight chlorides (EPA, 1987d). Recovery of beneficial chemicals is accomplished through combustion in a Kraft black liquor recovery furnace. The concentrated black liquor derived from the pulping process is sprayed into a furnace equipped with a heat recovery boiler. The bulk of the inorganic molten smelt that forms in the bottom of the furnace contains sodium carbonate and sodium sulfide in a ratio of about 3:1 (Someshwar and Pinkerton, 1992). The combustion gas is usually passed through an electrostatic precipitator that collects particulate matter prior to being vented out the stack. The particulate matter can be processed to further recover and recycle sodium sulfate.

10. Sewage Sludge Incineration: EPA has conducted stack emission testing for dioxin from sewage sludge incineration at three multiple-hearth sewage sludge incinerators (EPA, 1987d). The homologue profile shown in Figure 3-1 was developed from tests on these three incinerators. Multiple hearth incinerators are the dominant technology in use in the United States today for the incineration of sewage sludge.

11. Granular activated carbon regeneration furnaces: Granular activated carbon (GAC) is an adsorbent that is widely used in the control of pollutants in wastewater discharged from chemical and pharmaceutical industries, and in the treatment of finished drinking water at water treatment plants. Industrial manufacture of activated carbon is mostly obtained from the heat treatment of nut shells and coal under pyrolytic conditions (Buonicore, 1992). The properties of GAC make it ideal for adsorbing and controlling vaporous organic and inorganic chemicals entrained in combustion plasmas, as well as soluble organic contaminants in industrial effluents and drinking water. The high ratio of surface area to particle weight (e.g, 600 - 1600 m²/g), combined with the extremely small pore diameter of the particles (e.g., 15-25 Å) increases the adsorption characteristics (Buonicore, 1992). GAC will eventually become saturated and the adsorption properties will significantly degrade. When this occurs, the GAC usually must be replaced and discarded, which significantly increases the costs of pollution control. The introduction of carbon reactivation furnace technology in the mid 1980's created a method involving the thermal treatment of used GAC to thermolytically desorb the synthetic compounds and restore the adsorption properties for reuse (Lykins, et al., 1987).

The used GAC can contain compounds that are precursors to the formation of PCDD/Fs during the thermal treatment process. EPA measured precursor compounds in spent GAC used as a feed material to a carbon reactivation furnace tested during the National Dioxin Study (EPA, 1987d). The total chlorobenzene content of the GAC ranged from 150 ppb to 6,630 ppb. Trichlorobenzene was the most prevalent species present, with smaller quantities of di- and tetra-chorobenzenes detected. Total halogenated organics were measured to be about 150 ppm.

EPA has stack tested two GAC reactivation furnaces for the emission of dioxin (EPA, 1987d; Lykins, et al., 1987). The homologue profile shown in Figure 3-1 was developed from the tests at these two facilities. One facility was an industrial carbon reactivation plant, and the second facility was used to restore GAC at a municipal drinking water plant. The industrial carbon regeneration plant processed 36,000 kg/day of spent GAC used in the treatment of industrial wastewater effluents. Spent carbon was reactivated in a multiple-hearth furnace, cooled in a water quench, and stored and shipped back to primary chemical manufacturing facilities for reuse. The furnace fired natural gas, and consisted of seven hearths arranged vertically in series. The hearth temperatures ranged from 480° C to 1000° C. The spent GAC contained about 40% weight moisture. The used GAC was fed to the top hearth. In the furnace, the spent carbon was dried and the organics adsorbed onto the carbon particles were volatilized and burned in the heated combustion atmosphere. The regenerated carbon dropped from the bottom hearth of the furnace to a quench tank to reduce the temperature. Air pollutant emissions were controlled by an afterburner, a sodium spray cooler, and a fabric filter. Temperatures in the afterburner were about 930° C.

The second GAC reactivation facility tested by EPA consisted of a fluidized-bed furnace located at a municipal drinking water treatment plant (Lykins, et al., 1987). The furnace was divided into three sections: a combustion chamber, a reactivation section and a dryer section. The combustion section was fired by natural gas, and consisted of a stoichiometrically balanced stream of fuel and oxygen. These expanding gases of combustion provided heat, and suspended and fluidized the carbon. Temperatures of combustion were about 1,000° C. The reactivation section outside the combustion chamber allowed for the complete volatilization of the heated GAC. Off-gasses from the reactivation/combustion section were directed through an acid gas scrubber and high-temperature afterburner prior to discharge from a stack.

Another combustion process for which emissions data were sought was coal combustion in electric power generating facilities (utility boilers). Currently there is conflicting and extremely limited data on emission of dioxin-like compounds from coal-fired utility boilers (NATO,1988). The few published results of stack testing and monitoring of emissions from facilities in the United States have shown that dioxin has not been detected in stack gas emissions (NATO, 1988). Therefore, a homologue profile for this source category was not developed. The federal Clean Air Act requires an assessment of stack emissions of toxic air contaminants, including CDDs and CDFs, from coal-fired utility boilers. The EPA is currently collaborating with the U.S. Department of Energy in stack sampling seven facilities for CDD and CDF emissions. These results will be included in the final version of this document.

A homologue profile also could not be developed for a industrial category of potential concern, portland cement kilns. The database on stack emissions of dioxin from these units is just now becoming available. Volume 2, Chapter 3 of this assessment reviews current emissions inventories for cement kilns burning and not burning hazardous waste as auxiliary fuel in the production of cement clinker. In evaluating the TEQ of the mixture of CDDs and CDFs discharged from the stack of individual facilities, it became apparent that there was no consistent pattern to the relationship of total CDDs and CDFs to the estimated dioxin TEQ. For example, the ratio of total PCDD/Fs to the TEQ ranged from about a factor of 5:1 to a factor of 1000:1. A lower ratio reflects a skewing towards penta and tetra-chlorinated congeners in the distribution, and a higher ratio reflects a greater proportion of hexa, hepta, and octa chlorinated congeners in the emissions. Until more information becomes available from stack testing additional sources, a homologue profile of this industry will not be derived from the existing data.

The emission factors for the dioxin-like compounds from the stack of the hypothetical waste incinerator were derived from actual stack monitoring and emissions testing of an incinerator burning a complex mixture of organic waste. The concentrations of the specific PCDD/F congeners in units of nanograms per dry standard cubic meter (at 20° C; 1 atm.; 7% O₂) were available, as were the volume of gas escaping from the stack and feed rates for the material being combusted during the stack tests. Using procedures described in Section 3.2.1, this data was converted to emission factors. Such factors for three test runs are shown in Table 3-2. The fourth column is the average of these emission factors converted to g/sec units, which are the appropriate units for the application of the COMPDEP model. The conversion assumed a constant feed rate of 200 metric tons of feed material per day (further details on the hypothetical incinerator are found in Section 3.5). Human exposures to the coplanar PCBs emitted from a combustion source is not demonstrated in Chapter 5. Therefore, an estimation of congener-specific emission factors of coplanar PCBs for the hypothetical incinerator are not provided.

In order to put the emissions from the hypothetical waste incinerator into perspective, they can be compared with emissions from other incineration sources that are similarly controlled, e.g., equipped with scrubbers and/or fabric filters.

Such air pollution control devices can reduce the amount of dioxin that is formed within the system by >99% prior to the release from the stack. In this comparison, emissions from the following types of incineration processes were used (CARB, 1990; EPA, 1993): medical waste incineration; hazardous waste incineration; sewage sludge incineration; and municipal solid waste incineration. For comparisons, all emissions factors are expressed in units of nanograms TCDD-TEQ (Toxic Equivalent) emitted from the stack per kg of waste combusted, and are presented as ranges in measurements (minimum to maximum). This should not be confused as typical of the incineration source category, but specific only to sources having scrubbers and/or fabric filters. Volume 2, Chapter 3 of this assessment gives an overview of dioxin emissions from incineration technologies equipped with a variety of pollution control systems. The emissions from the hypothetical incinerator is ranked with the other types of waste incinerators that are well controlled with some combination of a scrubber device and/or a fabric filter, as follows:

1. Medical waste incineration: 25 - 200 ng TEQ/kg waste combusted.

2. Hazardous waste incineration: 0.18 - 119 ng TEQ/kg waste combusted.

3. Hypothetical waste incinerator: 4.5 ng TEQ/kg waste combusted. 4. Municipal solid waste incineration: 0.05 - 3 ng TEQ/kg waste combusted.

5. Sewage sludge incineration: 0.002 - 0.03 ng TEQ/kg sludge combusted.

From these comparisons it appears that the TCDD-TEQ emission factor derived for the hypothetical incinerator lies well within the range of emission factors developed from measured incineration sources burning a diversity of waste material, but employing similar air pollution control technology. The hypothetical incinerator was arbitrarily assigned a waste combustion rate of 200,000 kg waste/day. This charging rate conforms to a large medical waste incinerator, an average hazardous waste facility, and moderate sewage sludge and municipal waste incinerators.

The first step in the air modeling is the partitioning of total emissions into a vapor and a particle state. This section will review data on partitioning at the point of stack emission, in ambient air, and a theoretical approach to estimating the partitioning of dioxin-

Table 3-2. Emission factors and average emissions used for the hypothetical incinerator.

Emission Factors

Congener Test 1 Test 2 Test 3 Emissions

---------------- ng/kg ---------------- - g/sec -

2378-TCDD 0.052 0.031 0.037 9.3E-11

Other TCDD 0.826 0.870 0.913 2.0E-9

12378-PeCDD 0.148 0.056 0.048 1.9E-10

Other PeCDD 1.390 0.322 0.783 1.9E-9

123478-HxCDD 0.104 0.165 0.056 2.5E-10

123678-HxCDD 0.157 0.187 0.130 3.6E-10

123789-HxCDD 0.148 0.165 0.117 3.3E-9

Other HxCDD 2.440 0.670 1.040 3.2E-9

1234678-HpCDD 2.350 0.957 0.957 3.3E-9

Other HpCDD 4.040 1.650 2.170 6.0E-9

OCDD 4.260 1.390 3.130 6.7E-9

2378-TCDF 3.300 2.390 2.170 6.0E-9

Other TCDF 20.00 15.70 14.30 3.8E-8

12378-PeCDF 0.435 0.165 0.226 6.3E-10

23478-PeCDF 0.243 0.139 0.122 3.9E-10

Other PeCDF 6.280 4.480 3.480 1.1E-8

123478-HxCDF 0.478 0.365 0.357 9.2E-10

123678-HxCDF 0.478 0.343 0.313 8.7E-10

123789-HxCDF 0.357 0.165 0.226 5.7E-10

234678-HxCDF 0.243 0.117 0.074 3.3E-10

Other HxCDF 1.490 0.313 0.943 2.1E-9

1234678-HpCDF 0.243 0.565 0.696 1.2E-9

1234789-HpCDF 0.391 0.096 0.165 5.0E-10

Other HpCDF 241.0 2.380 2.180 5.4E-9

OCDF 1.570 0.478 0.971 2.2E-9

like compounds in ambient air. The true vapor/particle partitioning of dioxin under different conditions has not been directly measured, and therefore, is usually implied from these limited data or by theoretical means.