Envlron. Scl. Technol. 1084, 18, 171-179
Incineration of Polychlorinated Biphenyls in High-Efficiency Boilers: A Viable Disposal Option Gary 1. Hunt,' Paul Wolf,+ and Paul F. Fennelly
GCA/Technology Division, Bedford, Massachusetts 0 1730 Approximately 750 million pounds of polychlorinated biphenyls (PCBs) remain in service today in the United States. The eventual disposition of these materials and the vast stuck piles already removed from commerce and use represents a formidable problem to both U.S. industry (e.g., utility companies) and federal and state environmental agencies. Despite the fact that available disposal options include the use of high-temperature incineration, disposal efforts have been significantly hampered by the lack of approved incineration facilities. The results of comprehensive PCB incineration programs conducted in accordance with EPA test protocols at each of three high-efficiency boiler sites are presented. Flue gas sampling procedures included the use of both the modified method 5 PCB train and the Source Assessment Sampling System (SASS). Analytical protocols included the use of gas chromatography (GC/ECD) and combined gas chromatography/mass spectrometry (GC/MS). PCB destruction efficiency data for each of nine test runs were in excess of the 99.9% values assumed by the EPA regulations. The cumulative data set lends further credibility to the use of high-efficiency boilers as a viable disposal option for PCB contaminated (50-500 ppm) waste oils when conducted in strict accordance with existing EPA protocols. W
Introduction Industrial Use of PCBs. Polychlorinated biphenyls (PCBs) represent a class of synthetic chlorinated aromatics having unique physical and chemical properties. They are thermally stable, are resistant to degradation, are soluble in most common organic solvents, are fire resistant, and possess superior dielectric properties. These qualities have made PCBs an important industrial product since their introduction in the United States in 1929 until 1977 when manufacturing was halted. The superior dielectric properties of these materials, coupled with their thermal stability has prompted United States industry to adopt them for routine use primarily in electric transformers and capacitors (1,2). Recent estimates (3) indicate that approximately 1.4 billion pounds of these materials were purchased by American industry during the period spanning from 1929 to 1977. To date, an estimated 750 million pounds remain in service (450 million pounds in capacitors and 300 million pounds in transformers), 162 million pounds of which are associated with the electric utility industry (4). Disposal of PCBs. Traditionally, polychlorinated biphenyls have been prepared after treatment of a biphenyl feedstock with anhydrous chlorine in the presence of a catalyst such as iron filings or ferrous chloride. The resultant product is a complex mixture containing any number of 209 possible isomers. Government concerns over the suggested health implications of PCBs, their persistent bioaccumulative behavior, and long-term stability coupled with their widespread distribution and use have prompted strict controls over 'Present address: Metcalf & Eddy, Inc., Boston, MA 02114. 0013-936X/84/091&0171$01.50/0
the manufacturing, transport, and use of PCBs and PCBcontaining materials (2). PCBs were, in fact, the first chemical compound to be regulated by the Environmental Protection Agency (EPA) under the Toxic Substances Control Act (TSCA) of 1976. Regulations promulgated under this act prohibited further production of PCBs and established acceptable disposal technologies for PCBs and PCB-containing materials. The disposal technologies differ depending on the characteristics of the (solid or liquid) waste and the PCB concentrations. Materials containing in excess of 500 ppm of PCBs must be disposed of either by high-temperature incineration (for liquids) or in an approved chemical waste landfii (for solids). No disposal restrictions were imposed on materials containing less that 50 ppm of PCBs. For materials containing between 50 and 500 ppm of PCBs, several disposal options are acceptable, including approved incinerators, chemical waste landfills, and high-efficiency industrial and utility boilers (2). In addition, other methods may be used with the approval of the EPA Regional Administrator. Under this option, several nonthermal disposal processes have been approved. A partial listing of these surrogate disposal options includes alkali metal dehalogenation (5), solvent extraction ( 4 ) , and the plasma arc process (6). Despite specifications made in the EPA ruling (2) that PCBs stored for disposal prior to Jan 1, 1983, must be disposed of by Jan 1, 1984, control efforts have been hampered significantly by the lack of approved incineration facilities. As of this writing only three land-based commercial incineration facilities are licensed by the Environmental Protection Agency Energy Systems Company Facility in El Dorado, AR, the Rollins Environmental Services Company Facility in Deer Park, TX, and a General Electric facility in Pittsfield, MA. While there are several approved chemical waste landfills, they are often located at substantial distances from the industry or utility that must dispose of the PCB-contaminated fluid. Disposal costs, especially the transportation charge associated with disposal, can therefore be expensive. This situation has led to the emergence of high-efficiency boilers as an increasingly viable disposal option for PCB-contaminated fluids in the 50-500 ppm range. Virtually all industries and utility systems with significant amounts of PCBs own and operate boilers that can potentially qualify as disposal sites. Use of a company-owned boiler saves the cost of PCB disposal while recovering the heat content of the contaminated oil. This dual economic incentive has made the use of high-efficiency boilers an increasingly popular PCB disposal option. By the summer of 1982, 25 industrial and utility concerns had made contact with EPA regional offices concerning use of their high-efficiency boilers for disposal of PCB fluids in the 50-500 ppm concentration range. To date, 15 of these boilers have been given EPA approval to conduct a PCB test. Of these 15, 11 facilities are known to have burned PCB fluids, and 7 of these burns were monitored for PCB destruction efficiency. Cumulative results for three of these test programs are the subject of this presentation.
0 1984 American Chemical Society
Envkon. Scl. Technol., Vol. 18, No. 3, 1984
171
Research has indicated that a residence time of 2 s in a combustion zone maintained at 2000 O F in conjunction with an excess oxygen concentration of greater than 3% in the flue gas will ensure a PCB destruction efficiency in excess of 99.9% as assumed by the EPA protocols (7). High-efficiency boilers are categorized as boilers that operate with a combustion efficiency in excess of 99.9% as determined by the concentrations of CO and C02in the combustion zone. It is assumed that such boilers can attain a PCB thermal destruction efficiency in excess of 99.9% when operated under controlled conditions as specified in the regulations. At present, the promulgated regulations mandate not only PCB content in the fuel feed but also a list of boiler operating conditions including, but not limited to, combustion temperature, residence time in the boiler, and percent O2and CO (see Table I) in the flue gas. Additionally, it is well documented that both particulate and gaseous emissions from some municipal incinerators may contain measurable quantities of polychlorinated dibenzodioxins (PCDDs) and chlorinated dibenzofurans (PCDFs) (8-13). As a consequence, some concerns can be offered for the presence of these toxicants as potential combustion byproducts during the incineration of polychlorinated biphenyls in high-efficiency boilers. While these components have not typically been noted in controlled incineration process emissions such as high-efficiency boilers, numerous investigators have noted their presence under both controlled laboratory and actual field sampling activities (8,14-19). Buser et al. reported that significant concentrations of numerous PCDD and PCDF isomers were formed during the pyrolysis of an Aroclor mixture (16). Similarly, the unintentional combustion of capacitor or transformer fluids during an electrical fire resulted in the formation of significant concentrations of PCDD and PCDF isomers (18, 19). While these components are not typically measured during the controlled incineration of PCB-contaminated waste oil, the potential for their formation as combustion byproducts should be considered. The present investigation included a comprehensive evaluation of PCB incineration programs at each of three high-efficiency boilers located in the continental United States. Each of the sampling and analysis programs were designed to evaluate PCB destruction efficiency and verify that tests were in compliance with the existing EPA protocol governing the controlled incineration of PCB-contaminated oil in high-efficiency boilers. Continuous measurements for CO, 02,and C02 in the flue gas were conducted in the field. In addition, time-integrated samples of both the influent fuel feed (supplemented with PCB) and the corresponding flue gas emissions were collected tQ provide PCB destruction efficiency measurementa for each of the three units. Additionally both flue gas and particulate samples were collected for the analysis of polychlorinated dibenzofurans (PCDFs) and polychlorinated dibenzodioxins (PCDDs). Experimental Section Field Sampling. A variety of sample types were collected at each of three high-efficiencyincinerators as noted below. (1) Site A: Industrial Oil-Fired Boiler, General Motors, Bay City, MI. Tests were conducted in May, 1980, at a high-efficiency boiler owned and operated by the Chevrolet Motors Division of General Motors Corp. located in Bay City, MI. Tests were conducted on the no. 3 unit, Wickes type “K” 65-4K-7 package boiler with a rated 60 000 lb/h steam generation capacity based on an 8 gal/min feed rate of no. 6 fuel. Sampling of the incin172
Environ. Sci. Technoi., Vol. 18, No. 3, 1984
FLORlSlL
THERMOMETER
PRO
REVERSE-TYPE PITOT TUBE
THERMOMETER BY-PASS
VACUUM
(
ORIFICE MAIN VALVE ORIFICE MANOMETER
METER
AIR
TIGHT IMPINGERS I DEIONIZED, DISTILLED _ _ ~ WATER 2 - DEIONIZED
-
DISTILLED’ WATER
3 - DRY 4 - SILICA GEL
Figure 1. PCB sampling train modified method 5.
erator flue gas for PCBs was conducted on each of 3 separate days. Samples were typically collected over a 6-h period with the boiler operating at a normal load, burning no. 6 fuel oil supplemented with a contaminated waste oil containing approximately 500 ppm of Aroclor 1242 (9:l ratio). A preliminary test was conducted 1 day prior to the 3-day PCB test program to provide background emission data with the boiler burning no. 6 fuel oil only. Flue gas samples were collected on a Research Appliance Corp. (RAC) train which incorporated EPA approved modifications for the collection of PCBs. A schematic of the modified method 5 train is shown in Figure 1. Total elapsed sampling time on days 1-3 was 300 min (7 min per point) with a total gas volume of 5 m3 collected. Sampling on test day 4 was 210 min with a total sample volume of 3 m3 collected. As specified in the EPA protocols, a field blank train was set up and recovered on each of the 4 sampling days. As an additional quality control measure, duplicate samples were collected on test day 3 by using two separate trains operating simultaneously. Continuous monitoring measurements were provided for 02,C02,CO, and hydrocarbons on each of the 4 test days. Each of these monitors were calibrated twice daily by using certified (*1%) reference gases obtained from Scott Environmental Technology. Oxygen concentrations were determined on a Beckman Model 741 paramagnetic O2 analyzer with a measuring range of &lo% O2full scale. The analyzer was calibrated at 0% O2 with ultrapure nitrogen and with 8.03% O2 certified calibration gas before and after each test. Carbon dioxide concentrations were determined on an Infrared Industries Model 702 NDIR carbon dioxide analyzer with a measuring range of 0-30% COPfull scale. The analyzer was calibrated at 0% C02 with ultrapure nitrogen and with 7.99% C02 certified calibration gas before and after each test period. Carbon monoxide concentrations were determined on a Beckman Model 65 NDIR CO analyzer with a measuring range of 0-50 ppm of CO full scale and was calibrated at 0 ppm of CO with ultrapure nitrogen and with 39.9 ppm of certified calibration gas before and after each test period. All sampling equipment was cleaned and prepared at the GCA facility in Bedford, MA, as detailed earlier (20). A representative sample of the contaminated waste oil was collected during each of the verification tests. These were extracted from the fuel oil line during each test. (2) Site B: Utility Coal-Fired Boiler, Union Electric, St. Louis, MO. Tests were conducted in January
of 1981 at a boiler owned and operated by the Union Electric Co. of St. Louis, MO. Tests were conducted in a coal-fired unit located at the Labadie power station. The no. 4 unit selected for testing consisted of a Combustion Engineering tangentially fired boiler with a rated input capacity of 5400 X lo6 Btu/h. Sampling of the boiler flue gas for PCBs was again conducted on each of 3 separate days. Stack samples were typically collected over a 9-h period with the boiler operating at a normal load, burning pulverized coal supplemented with contaminated waste oil containing 100 ppm of Arochlor 1260. As before a preliminary background test was conducted prior to the 3-day test program with the boiler burning coal only. Flue gas samples were collected by using the modified collection system shown in Figure 1 (21-23). Sample volumes on each of the test days were typically 10-12 dscm (dry standard cubic meters). As noted earlier field blank trains were set up and recovered on each of the 4 sampling days. All sampling equipment and associated containers were cleaned and prepared as described earlier (20). Continuous monitoring measurements were again provided for 0 2 , C02, and CO on each of the 4 test days. Oxygen concentrations were determined on a Horiba Model POA 21 polargraphic O2analyzer with a measuring range of 0-20% O2full scale. Carbon dioxide concentrations were determined on a Horiba Model PIR 2000 NDIR carbon dioxide analyzer with a measuring range of 0-25% nitrogen. Carbon monoxide concentrations were determined on a Horiba Model PIR 2000 NDIR CO analyzer with a measuring range of 0-500 ppm of CO full scale. Representative composite samples of the contaminated waste feed and fly ash from the particulate control device were collected during each of the four stack tests. (3) Site C: Utility Oil-Fired Boiler, Northeast Utilities, Middleton, CT. Testa were conducted in Sept, 1981, at a boiler owned and operated by Northeast Utilities of Hartford, CT. The tests were conducted in the no. 3 boiler in their Middletown, CT, facility. The no. 3 boiler is a 2185 X lo6 Btu/h boiler, rated at 233 MW. The unit was normally burning no. 6 fuel oil. Four days of testing were conducted. These 4 days consisted of 1-day burning an uncontaminated no. 6 oil as a surrogate to the contaminated oil. On the 3 subsequent days, contaminated fuel was pumped in an analogous fashion as the no. 6 oil. In each case approximately 900 gal/h was input into the boiler. This input was less than 10% of the total fuel feed. The contaminated fuel was analyzed to have a concentration of 200-410 ppm of Aroclor 1260. Stack samples were collected over an 8-h period with the boiler operating at normal load. Flue gas samples were collected by using both the modified collection system shown in Figure 1 (21-23) and the higher volume Source Assessment Sampling System (SASS)shown in Figure 2. The sample volumes on each of the days using the modified system were nominally 8 dscm, whereas the SASS train collected 50 dscm. As noted earlier, field blanks for the modified PCB train were set up and recovered on each of the 4 test days, while blank SASS trains were set up and recovered on 2 of the 4 days. All sampling equipment and associated containers were cleaned and prepared as described earlier (20). Continuous monitoring was performed for O2 and CO on each of the 4 test days. Oxygen concentrations were determined on a Hartmann & Braun Magnos 5 magnetic oxygen analyzer. Carbon monoxide concentrations were determined on a Bendix 8501-5CA NDIR CO analyzer.
,1STACK
CONVECTION
FILTER
TC.
I
GAS
._ PITOT
n
COOLER
I
W
Figure 2. Source Assessment Sampling System (SASS): Utility oiCfired boiler, site C.
Representative composite samples of the contaminated waste feed and no. 6 feed were collected on each day that the respective fuel was fed into the boiler. Laboratory Analyses. (1) Site A: Industrial OilFired Boiler, General Motors, Bay City, MI. Flue gas samples and representative samples of the fuel feed from each of the test burns were returned to the GCA laboratory for analyses. A general analytical flow scheme for the flue gas samples is depicted in Figure 3. As shown, each of the sampling trains (see Figure 1)consisted of a Florisilsorbent cartridge, a series of aqueous impingers, and hexane and acetone train rinses. The contents of each of the Florisil-adsorbent cartridges were Soxhlet extracted with 200 mL of hexane for 4 h. Aqueous impinger samples were combined in the field and returned to the laboratory as a single sample. The combined sample was extracted with three 100-mL portions of hexane and combined with the Florisil concentrate. The hexane and acetone rinses were concentrated and combined with the impinger water and Florisil-sorbent concentrates. The resultant extract was dried with a sodium sulfate column and concentrated under reduced pressure with a rotary evaporator at 40 OC. The combined extract was reduced to approximately 10 mL and partitioned against concentrated sulfuric acid in a separatory funnel. The acid layer was discarded, and the organic layer was restored to 10 mL. A 1-mL aliquot was removed for prescreening by using a gas chromatograph fitted with an electron capture detector (63Ni). All analyses were conducted on a Hewlett-Packard 5840A gas chromatograph. Pertinent instrument operating parameters were as follows: column 6 f t X 2 mm (i.d.) 1.5% OV-17/1.95% QF-1 on 100/120 Chromosorb WHP, column temperature 175 "C, injector temperature 270 OC, detector temperature 300 "C, and carrier flow 50 mL/min argon/methane (95/5). Instrument calibration was accomplished with an Aroclor 1242 reference standard distributed by Applied Science Labs, Inc., State College, PA. Details of PCB quantitation and calibration are provided in an earlier publication (20). Additionally two 3-mL aliquots were removed for PCB isomer speciation by using combined gas chromatography/mass spectrometry employing a Hewlett-Packard GC/MS/DS system operating in the selected ion mode (SIM). Each of the duplicate aliquots was spiked, with anthracene-dlo serving as a quantitative internal standard. Analyses were conducted in the selected ion mode for a variety of individual PCB positional isomers. Pertinent instrument operating conditions were as follows: column 6 f t X 2 mm (i.d.), 1% SP2250 on 80/100 Supelcoport, column temperature 160 (4min)-270 OC at 10 OC/min and held for 15 min, injector temperature 275 "C, scan time 250 ms/amu, and electron energy 70 eV. Instrument Environ. Sci. Technoi,, Vol. 18, No. 3, 1984
173
IMPINGER WATERS
.1
COMBINED TRAIN RINSES (HEX~NEIACETONE)
EXTRACT (WIHEXANE)
FLORlSlL
ADSORBENT
4
TUBE
EXTRACT ( W I HEXANE)
\1 CONCENTRATE
3.
CONCENT RATE
CONCENTRATE
J;
coMrNE
CONCENTRATE ( I O ml) PARTITION
J. J.
CLEANUP ( H z S 0 4 )
GC/ ECO PRE-SCREENING
J
FURTHER CLEANUP REOUIRED
COLUMN CLEANUP
I
..
NO GC/ECD (AROCLOR SPECIFIC
.1RESPONSE PATTERN, ISOMER)
$YES CONCENTRATE ( I F NECESSARY)
.1
GC/ MS SELECTED ION MON ITOR ING
Figure 3. General analysis scheme: PCBs in flue gas.
calibration was provided by using a series of representative PCB positional isomers supplied by RFR, Inc., Hope, RI, including biphenyl, 2-chlorobiphenyl,3,3'-dichlorobiphenyl, 2,4,5-trichlorobiphenyl,2,3,5'-trichlorobiphenyl, 2,3',4',5tetrachlorobiphenyl, 2,2',4,5,5'-pentachlorobiphenyl, 2,2',4,4',6,6'-hexachlorobiphenyl, 2,2',3,4,5,5',6-heptachlorobiphenyl, and decachlorobiphenyl. GC/MS response factors relative to anthracene-& were provided for three concentrations of each of these positional isomers. Fuel oil analyses consisted of duplicate analyses of a representative fuel feed sample from each of the test days. Initially, sample bottles were heated at 40 "C in a water bath and stirred for 1 min prior to aliquoting. A 0.5-g portion of each sample was transferred to a 5.0-mL volumetric flask and diluted to the mark with hexane. The resultant sample was shaken with 5.0 mL of concentrated sulfuric acid and the oil phase removed and treated with 5.0 mL of 10% NaHC03. An aliquot of the oil fraction was then removed for PCB analyses by using a gas chromatograph as described earlier. Details of the instrument calibration procedure and qualitative/quantitative analyses techniques are provided in an earlier publication (20). Quality control protocols included the analyses of method blanks, field blanks, and both Florisil-sorbent and fuel oil spikes containing predetermined quantities of hoclor 1242. (2) Site B: Utility Coal-Fired Boiler, Union Electric,'St. Louis, MO. -Flue gas samples were received from each of four test burns. These were prepared in an identical manner to the site A samples with the exception that the combined impinger/Florisil rinse concentrate was reduced to 1mL and adjusted to 2.0 mL prior to GC/ECD analysis. A 0.5-mL aliquot was removed for PCB screening employing GC/ECD. Instrument operating conditions were as described earlier with the following exception: column temperature 175 (2 min)-185 "C at 2 OC/min and carrier flow 40 mL/min argon/methane (95/5). The presence of significant levels of electron-capturing interferences indicated the need for acid partitioning. As a 174
Environ. Sci. Technoi., Vol. 18, No. 3, 1984
consequence a 0.5-mL portion of each 2.0-mL extract was diluted to 1.0 mL and partitioned with concentrated sulfuric acid. The resultant extracts were reanalyzed, and the interferences were sufficiently reduced to permit interpretation. Instrument calibration was accomplished with an Aroclor 1260 standard reference material procured from the EPA reference repository in Research Triangle Park, NC. Further confirmation of the presence of specific PCB isomers in each sample was provided by using gas chromatography/mass spectrometry. Analyses were conducted on a Hewlett-Packard 5985 GC/MS/DS system operated in the selected ion mode (SIM). Pertinent operating parameters were as follows: column 30-m SE-54 fused silica capillary wall coated, column temperature 50 "C (2 min) at 10 "C/min to 260 "C and held, splitless injection (1pL), injection temperature 275 "C, sweep time 30 s, electron energy 70 eV, and scan time 250 ms/scan. Instrument calibration was provided as before (site A) by using a variety of PCB positional isomers obtained from RFR, Inc., Hope, RI. GC/MS response factors were again provided relative to the internal standard anthracene-& Waste oil samples, composited daily during each test burn, were aliquoted in duplicate for GC/ECD analysis. A 1-g portion of oil was diluted with hexane to 10 mL. Each sample was partitioned with sulfuric acid and a 1-mL portion removed for GC/ECD analysis. Instrument operating conditions and calibration procedures were identical with those used for analyses of each of the flue gas samples. Quality control protocols specific to this test program included the following: method blanks, field blanks, and Florisil-sorbent and waste oil samples fortified with predetermined quantities of Aroclor 1260. Additionally EPA/EMSL quality control concentrates of Aroclor 1260 were analyzed to verify the accuracy of the Aroclor 1260 calibration standard. (Replicate analysis of a 0.50 pg/mL Aroclor 1260 EPA concentrate using available calibration curves yielded values of 0.47 and 0.54 pg/mL.)
(3) Site C: Utility Oil-Fired Boiler, Northeast Utilities, Middletown, CT. Flue gas samples were again collected during each of four test burns. Method 5 train (PCB) samples were prepared in an identical fashion to the site A and B samples noted earlier. Additionally each of the Florisil-adsorbent tubes were spiked with anthracene-dlo(surrogate component) prior to the extraction step. The combined extracts (Florisil, impinger, and train rinses) were concentrated as before and restored to 3.0 mL for subsequent analyses. A 0.5-mL aliquot was removed for the gas chromatographic (GC/ECD) prescreening procedure. Instrument operating conditions were as described earlier (site A) with the following exception: carrier flow 40 mL/min argon/methane (95/5). The presence of significant levels of electron-capturing interferences necessitated the use of an acid-partitioning procedure. As a consequence, a second 0.5-mL aliquot was removed from each sample, adjusted to 1.0 mL, and partitioned with concentrated sulfuric acid. Analyses proceeded by employing the GC/ECD protocols described earlier. Results, again indicated the persistence of high levels of background contamination in each of the test burn samples. This contamination persisted in the field blank train, also, but was not in evidence in laboratory method blanks or fieldbiased blank reagents. In an attempt to remedy this situation, subsequent cleanup proceeded by employing alumina column chromatography (Figure 3). The details of this technique are provided elsewhere (20). Further gas chromatographic analyses were again unsuccessful, and as a consequence 1.0-mL aliquots of each extract were reduced to 0.1 mL for GC/MS analysis. This particular step included transferring each extract to a calibrated centrifuge tube with subsequent treatment using a gentle stream of prepurified nitrogen to reduce the extract volume to 100 pL. GC/MS analyses were again conducted on a Hewlett-Packard 5985 GC/MS/DS system operated in the selected ion mode (SIM). Pertinent instrument operating conditions were as follows: column 6 ft X 2 mm (i.d.) 1.5% OV-17/1.95% QF-1 on 100/120 Chromosorb WHP, column temperature 160 OC (2 min) at 5 OC/min to 225 OC and held, injector temperature 225 "C, injection volume 3 pL splitless, scan time 0.8 s/scan, and electron energy 70 eV. Instrument calibration was again provided by using a variety of PCB positional isomers obtained from RFR, Inc., Hope, RI. GC/MS response factors relative to naphthalene-d, were provided for three concentrations of each of the positional isomers noted earlier. Additionally, flue gas samples collected by using the Source Assessment Sampling System (SASS) were returned to the laboratory for analysis. As noted earlier, a SASS train was operated simultaneouslywith the modified method 5 (PCB) train during each of the four test runs. As noted in Figure 2, the SASS configuration included a particulate filter, 1-pm cyclone, a series of aqueous impingers, and an organic-sorbent module containing approximately 150 g of XAD-2 (copolymer of styrene-divinylbenzene). XAD samples were surrogated spiked with anthracene-dIo (19 pg) and Soxhlet extracted with methylene chloride for 24 h. The solvent extract was then held for combination with the impinger extracts and train rinse. The impinger waters were extracted in a separatory funnel with three 10% (v/v) portions of methylene chloride and the extracts combined with the appropriate XAD extract and train rinse. The combined extract was then dried with anhydrous sodium sulfate and concentrated to approximately 1mL with a Kuderna-Danish evaporator. Each concentrate was solvent exchanged into hexane and the volume adjusted to 3.0 mL. Procedures followed for
aliquoting the extracts, acid (H2S04)partitioning alumina chromatography, and subsequent gas chromatographic (GC/ECD) analyses were identical with those used for the method 5 train sample extracts. GC/ECD analyses were again inconclusive due to the presence of electron-capturing interferences persistent after cleanup procedures. Again analyses proceeded by using gas chromatography/mass spectrometry operated in the selected ion mode (SIM). As before, a 1.0-mL aliquot of each SASS train extract was reduced to 0.1 mL (100 pL) by using a gentle stream of prepurified nitrogen prior to analysis. GC/MS analyses were again conducted on a Hewlett-Packard 5985 GC/MS/DS system with operating conditions identical with those employed during analyses of the modified method 5 train samples. Instrument calibration procedures again employed a series of representative PCB positional isomers identical with those noted earlier. GC/MS response factors relative to naphthalene-& were provided for each of the positional isomers noted. Fuel oil analyses again consisted of duplicate analyses of a representative (composite)fuel feed sample from each of the 4 test days. A 1-g aliquot of each oil sample was transferred to a 10.0-mL volumetric flask and diluted to the mark with hexane. The resultant sample was partitioned against concentrated sulfuric acid and a 1-mL portion removed for subsequent GC/ECD analyses. Pertinent operating conditions were identical with those employed for analyses of the flue gas samples noted earlier. Instrument calibration was provided by using five serial dilutions of an Aroclor 1260 standard reference material obtained from the EPA repository in Research Triangle Park, NC. Calibration curves were constructed by performing a linear regression analysis on a series of five dilutions of the Aroclor 1260 reference standard. (The resultant correlation coefficient in all instances was in excess of 0.999.) Quality control protocols included the analysis of method blanks, field blanks, and both XAD-2 and Florisilsorbent spikes containing predetermined quantities of a series of chlorobiphenyl isomers. (The results of these analyses are provided in a subsequent portion of this paper.) As noted earlier each of the flue gas train samples was fortified with 10 pg of anthracene-dlo prior to the extraction sequence serving as a surrogate standard. (Surrogate recoveries for the modified method 5 train samples averaged 66% (n = 4) during the course of the test program, while the SASS train samples averaged 67 % (n = 4).) Quality control measures for the fuel oil analyses consisted of the analyses of duplicate aliquots of an EPA/ EMSL PCB in transformer oil check sample processed simultaneously with the actual program fuel samples. Results are, again, provided in a subsequent section of this paper. Additionally, an EPA/EMSL check sample of Aroclor 1260 in acetone was diluted in duplicate and analyzed as a verification of the calibration standard. The observed concentration of this diluted sample was 97% of the expected concentration. Results
Test Measurements. Each of the three boilers selected for testing had historically demonstrated operating parameters in compliance with Federal specifications for PCB incineration in high-efficiency boilers (2). A summary of these pertinent criteria are shown in Table I. As shown in Table 11, the three units encompass a range of highefficiency boilers with rated energy inputs from 70 X lo6 Envlron. Sci. Technol., Vol. 18, No. 3, 1984
175
Table I. Federal Specifications for PCB Destruction in High-Efficiency Boilers ( 2 ) waste oil contains 50-500 ppm of PCB boiler is rated at minimum of 50 million Btu/h CO is less than 50 ppm (oil and gas) and. 100 ppm (coal) 0, in flue gas is greater than 3% waste oil is 10% or less (v/v) of fuel feed rate Table 11. High-Efficiency Boilers Pertinent Operating Parameters rated energy input, Btu/h boiler type
fuel
site A industrial site B utility site C utilitv
oil coal oil
lo6)
(X
70 5400 2185
waste fuel flow, gal/h
stack volume, dscm /h ( X 103)
19-20 500-580 907-915
18.9 2000-2200 750-775
to 5400 X lo6 Btu/h. Primary fuels ranged from no. 6 residual oil to coal. A summary of test results for the General Motors industrial boiler is shown in Table 111. This includes continuous monitoring values for O2 and CO as well as the resultant PCB destruction efficiency (DE) measurements. The promulgated EPA regulations specify a value of greater than 99.9% for the latter measurement as derived from the following general relationship: % destruction efficiency (DE) = PCB~ P~ C B ~ " ~ x 100 (1) PCBIN where PCBIN = input total PCB (grams) per test (Tables
111-V) and PCBouT = total PCB stack emission (grams) per test (Tables 111-V). A summary of test results for the Union Electric utility boiler is shown in Table IV. Again, all pertinent EPA specifications for the controlled incineration of PCBs in coal-fired boilers (see Table 11) were satisfied for each of the 3 test days noted. Destruction efficiency (DE) calculations again exceed the 99.9% values assumed by the EPA regulations. Test results for the Northeast Utilities Middletown no. 3 unit are shown in Table V. Destruction efficiency calculations are provided for flue gas samples collected by using both the modified method 5 train and Source Assessment Sampling System (SASS). While the SASS DE values were typically higher than the method 5 values, due solely to greater flue gas sample volumes (providing lower detection limits), both were in excess of the 99.9% values typically assumed by the EPA regulations. Quality Control Results. Quality control protocols for the three test programs collectively included the following: replicate flue gas samples, replicate analyses of waste fuel feed, analysis of background waste fuel supplemented with predetermined quantities of appropriate Aroclor mixtures, and replicate analyses of flue gas sorbents supplemented with specified quantities of PCBs and related isomers. While data on the actual collection efficiency of the respective sampling systems were not collected during any of the test programs, other investigators have reported the successful collection of a number of chlorobiphenyl isomers using a prototype to the method 5 train specified here (25). Our activities, however, did include percent recovery measurements for both Aroclor mixtures and specific chlorinated biphenyl isomers as applied to sorbent samples
Table 111. Summary of Results: Site A Industrial Oil-Fired Boiler, General Motors, Bay City, MI
test day background 1
%
[PCBl,b*eppm, in waste oil
total input PCB, g per test
14.2
6.3
-
8.7 9.6 9.6
6.1 5.6 5.6
340-720 340-760 340-760
[COI,' PPm
[O,l,"
total PCB stack emissions per testC g
1g/m3
PCB destruction efficiency, %
-
