PAGE 2 of 3

3.2.4.1. Vapor phase/particulate phase inferences from stack measurements

While the available literature is weak in this area, various investigators have made inferences on the vapor phase/particulate phase (V/P) partitioning from in-the-stack sampling of PCDD/F emissions from combustion sources. Sampling systems which have been used basically consist of a particulate filter followed by a section designed to condense vapors in impinger glassware surrounded by an ice bath, and a resinous material suitable for absorbing vapor phase compounds.

Depending on where the congener is distributed within the component parts of the sampling apparatus, the investigator reports the fraction associated with particulate, and the fraction found in the vapor absorbing material. In order to collect sufficient mass of particulate for accurate analytical determination of the concentration of the recovered congener at sub-part per trillion levels of detection, it is often necessary to sample in stack for periods of four hours or longer.

This introduces the possibility of movement of the collected dioxin sample from one part of the sampling train to another through adsorption, desorption, particulate blow-off, or other such phenomena as the sampling train continues to be exposed to the hot combustion gases. No real-time sampling method currently exists to instantaneously measure the concentration and physical state of the various PCDD/F congeners in the fluid turbulence of the hot combustion plasma characteristic of gases from combustion traveling up a cylindrical stack.

For these reasons, V/P partitioning based on stack test data is highly uncertain. Additional laboratory research is needed that is specifically directed at identifying the physical state partitioning of individual PCDD/F congeners at the exit to the stack under varying temperature profiles and conditions of particulate loading and acid gas concentration. Table 3-3 is a summary of the percent distribution of PCDD/Fs between the vapor-phase (V) and the particulate phase (P) as interpreted by various stack sampling techniques employed in the measurement of the compounds during incinerator operations.

table Table 3-3. Percent distribution of dioxins and furans between vapor-phase (V) and particulate-phase (P) as interpreted by various stack sampling methods.
Cavallaro, et al. (1982) performed a series of stack tests on six municipal solid waste (MSW) incinerators in Italy. He was one of the first investigators to interpret the V/P ratio from where the PCDD/F segregated with the sampling train, e.g., the particulate filter and resinous trap.

From these data, the percent distribution of congener groups were estimated. Cavallaro observed that the PCDD/F emissions from the stack of the tested incinerators seemed to predominate in vapor phase. He attributed this to the possibility that the relatively high temperatures of the combustion gases during sampling (700 to
900° C) may have promoted desorption of PCDD/Fs from particulate, although the sampling probe was kept at a constant 150° C.
expand table Table V3 3-3
  • Benfenati, et al. (1986)
    describes the stack testing of a modular MSW incinerator in Italy having a combustion capacity of 1500 kg/hour. The purposes of the study were to analyze the concentration of TCDD and TCDF at various points of the incineration process, to estimate the vapor phase versus the particulate phase partitioning at various sampling points corresponding to changes in temperatures, and to estimate the TCDD/TCDF control efficiency of the pollution control device (an electrostatic precipitator). Comparisons were made between the distribution of TCDD/TCDF after the secondary furnace in a region where combustion gas temperature was about 330° C, and the distribution at the stack where combustion gas temperature was 230° C. Benfenati observed that approximately 85% of the TCDD was in the vapor phase at the exit to the furnace, and approximately 95% of the TCDD was in the vapor phase at the stack. It was concluded that most of the TCDD predominated in vapor phase at the point of release from the stack at the reported temperature of 230° C. However, Benfenati could not exclude the possibility that the TCDD was adsorbed onto ultra fine, submicron aerosol particles.


  • Tiernan, et al. (1984)
    reported on the distribution of PCDD/Fs recovered in the stack sampling apparatus (EPA Modified Method 5) following the stack testing of a mass burn MSW incinerator operating in Japan. In the Modified Method 5 procedure, the sampling probe is maintained at a temperature of 120° C while the stack gases are isokinetically sampled. The facility was equipped with a dry scrubber combined with a fabric filter as the primary pollution control device. Tiernan observed congener-specific variability in the V/P partitioning inferred from the sampling method. However, greater than 55% of the PCDD/Fs were estimated to be in vapor phase at the point of release to the stack. In an earlier stack test (Tiernan, et al., 1982) of an MSW incinerator equipped with an electrostatic precipitator, Tiernan found that 45% to 89% of the PCDD/Fs were associated with particulate.


  • Clement, et al. (1985)
    stack tested a mass burn MSW incinerator operational in Canada for the emission of PCDD/Fs. Three 24-hour stack samples were taken using the EPA Modified Method 5 train with a stack temperature of 230 - 250° C. The components of the sampling train were analyzed separately. Clement observed that more than 95% of the total PCDD/Fs detected in the sampling train samples was found in the impingers used to condense vapor phase organic pollutants. Interpretation of this is difficult. However, it is implied from these data that most of the PCDD/Fs prevailed in vapor phase.

  • Hagenmaier, et. al (1986)
    conducted field tests of two different stack test methods for the accuracy, precision, and comparability of PCDD/F measurements. Both instruments were similarly constructed with a glass fiber filter for the capture of particulate-bound contaminants, a series of water or ice-cooled impingers to promote condensation of vapor phase contaminants, followed by an absorbing material to trap vapor phase pollutants. Eight parallel stack sampling experiments were carried out over a three week period using the sampling trains known as the German simple dilution method and the EPA Modified Method 5.

Although the two methods reported quite similar total concentrations of PCDD/Fs, the distribution of PCDD/Fs between the heated glass filter, and ice-cooled impingers and the sorbent trap were remarkably different. In one train, referred to as Train A by Hagenmaier, the temperature in the filter housing was 140° C, and in the second train, Train B, the temperature was 90° C. The stack gas temperature in both cases was 230° C. Hagenmaier found that the percentage of PCDD/Fs in the glass fiber filter was markedly greater in Train B than in Train A.

Up to 93% of the PCDDs and 90% of the PCDFs were detected in the particulate filter in Train B. By comparison, 73% and 58% of PCDDs and PCDFs, respectively, were detected in the particulate filter in Train A. Although Hagenmaier's data is used in Table 3-3, Hagenmaier theorized that this difference in the distribution of PCDD/Fs in the two sampling trains was due to the differences in the temperature of the glass fiber filter housing.

  • EPA (1990a)
    conducted a field validation study for the EPA stack testing Method 23 (the Modified Method 5) for the collection and retention efficiency of PCDD/Fs. A carbon-13 labelled congener was metered into the sampling probe just preceding the glass fiber filter using a dynamic spiking system. The validation procedure involved the isokinetic sampling in the stack of a large mass burn MSW incinerator. Sampling in situ in the stack while using a dynamic spiking system demonstrated that most of the isotope was recovered in the filter trap and front half of the sampling train designed to capture particulate, and a lower amount was recovered in the XAD resin designed to capture vapor phase organic compounds. In the particular tests in which the overall percent recovery of the dynamic spike were found to be acceptable, the XAD resin and condensor contained about 49% of the isotope, and 51% was associated with carbonaceous particulate.
3.2.4.2. Discussion of vapor/particle ratios derived from stack testing methods

It is apparent that the stack sampling method gives inconclusive and contradictory evidence of the V/P partitioning of PCDD/Fs at the stack of incinerators. Although most of the researchers report finding the greatest quantity of PCDD/Fs captured within the resinous material having the physical/chemical properties of absorbing vapor phase organic compounds, a few studies have reported the opposite. What is unusual about the V/P distributions in Table 3-3 is the lack of complete consistency despite the similarity of sampling method. Although the stack gas temperatures may vary, the probe and housing to the sampling instrument is usually kept at a standard temperature while traversing the hot flue gas. A more consistent pattern of V/P should have emerged.

Hagenmaier, et al. (1986)
has postulated that, depending on the temperature of the glass fiber particulate filter housing, the PCDD/Fs might desorb (volatilize) from particulate matter trapped in the filter during the 4 hours of sampling time required of the stack sampling method. Therefore, Hagenmaier does not believe that the distribution of PCDD/Fs between the particulate filter, the condensing impingers, and the vapor absorber gives a true indication of the V/P partitioning of these compounds at the stack.

Tests also have been devised by the EPA (1990a) to study the effect a change in temperature of the glass fiber filter housing might have on the distribution of PCDD/Fs in the sampling train. During the sampling period, two sampling trains were used: one inlet to the electrostatic precipitator (ESP), and the other placed near the outlet to the ESP. Temperatures of the filter housing were raised from the standard 120 C to 215 C in both sampling trains.

In agreement with the observations of Hagenmaier, et al. (1986), an increase in temperature generally resulted in a change in the distribution of the recovered 13-C labelled PCDD/F congeners. However, the temperature effect was most apparent within the sampling train inlet to the ESP. In the inlet sampling train, the higher filter box temperature increased the relative percentage of PCDD/Fs trapped in the impingers and XAD-2 resin.

An amount estimated to be in the vapor phase, based on the segregation of the compounds within the component parts of inlet sampling train, is as follows (with a range listed from low to high temperature):

TCDD = 20 - 55% vapor; HxCDD = 10 - 30% vapor; OCDD = 5 - 18% vapor; HxCDF = 18 - 58% vapor; OCDF = 5 - 18% vapor.

In the outlet sampling train (characteristic of stack emissions), this dramatic shifting of the congeners from the filter to the XAD-2 did not occur with an increase in temperature.

Interpretation of the vapor phase partitioning in the outlet sampling train from low to high temperatures was as follows:

TCDD = 90 - 95% vapor; HxCDD = 85 - 90% vapor; HxCDF = 90 - 95% vapor; OCDD = 75 - 90% vapor; OCDF = 78 - 90% vapor.

Both these interpretations were developed using a 500 ng PCDD/F spiked congener. Notice that the vapor phase to particle phase ratio is significantly different between the inlet and outlet sampling trains: in the inlet train most of the PCDD/F congeners seemed to predominate in the particle phase at the standard temperature of the filter housing, whereas in the outlet train most of the PCDD/F congeners seemed to predominate in the vapor phase, as interpreted by the distribution within the apparatus.

The temperature-dependent partitioning has recently been observed by Janssens, et al. (1992) during field validation studies involving the sampling of operating incinerators in Belgium. Janssens observed that the fraction of PCDD/Fs collected in the heated portion of the particulate glass filter (temperatures in the range of 250 to 300 C) showed an expected partitioning according to the vapor pressures of the compounds. It was found that a very low proportion of the PCDD/Fs were found in the particle phase; nearly all the compounds were detected in the vapor phase.

Moreover, Janssens observed that higher temperatures seemed to favor the vaporous state of the lower chlorinated congeners (compounds having one to five chlorines on the aromatic ring), and the particulate phase for higher chlorinated congeners (five to eight chlorines). This agrees well with the decrease in vapor pressures that occurs with an increase in chlorination, and an increase in vapor pressure that occurs with a decrease in chlorination of PCDD/Fs.

Adding to the theory of Hagenmaier, et al. (1986), Janssens believed that either the sampling apparatus was giving a true distribution of the V/P ratio of individual congeners, or that a significant portion of the congeners were reversibly sorbed onto particulate surfaces and could be eluded to vapor phase by the passage of the volume of sampled combustion gas over a lengthy time interval, neither of which could be proven by his study.

Benfenati, et al. (1986)
has suggested that what may be reported as vapor phase may actually consist of nucleated aerosol particles having diameters less than 0.1 micrometers. The impingers in the sampling method are located a few centimeters behind the heated particulate glass fiber filter, and are bathed in an ice bath. The dramatic reduction in temperature within the impinger glassware may cause sublimation from vapor phase to nucleation of aerosol particles. Downstream of the impingers is the vapor absorbing material, usually XAD-2 resin. Although this has been shown to be an excellent trap for semi-volatile organic compounds, the retention of submicron size particles with PCDD/Fs adsorbed onto the surfaces, or absorbed into the interior spaces, cannot be ruled out or excluded as a possible explanation for investigators reporting a preponderance of concentration both in the impingers and the vapor trap.

Complicating any meaningful interpretation of the data is the long duration of sampling time required in the stack measurement method. In order to reach a sub-ppt level of detection of PCDD/Fs for reliable quantification of specific congeners, sampling proceeds until approximately a five gram mass of particulate is gathered in the particulate filter. This may require in situ placement of the sampling apparatus such that samples are taken isokinetically, and the stack interior diameter is traversed for four or more hours.

Thus the sampling instrument is continuously exposed to the hot gas plasma over a long sampling moment. In addition the hot gases also contain precursor compounds, chlorides, oxides of sulfur and HCl which may have an effect on the success of accurately sampling PCDD/Fs. Although Janssens, et al. (1992), Hagenmaier, et al. (1986), and EPA (1990a) have all but excluded the possibility that sampling under these conditions creates results by producing PCDD/Fs or destroying PCDD/Fs somewhere within the sampling train, the possibility that the method creates an illusion of the true V/P ratio cannot be excluded.

The above discussions have indicated the variability in the data and the uncertainty with the stack results of vapor/particle partitioning. For these reasons, these data will not be used to infer the V/P distribution of PCDD/Fs at the point of release from the stack.

3.2.4.3. Vapor/particle partitioning of PCDD/Fs from ambient air sampling

The measurement of PCDD/Fs in air under ambient conditions has only been achieved since the late 1980's. Although there may be some variations, the ambient air sampler usually consists of a glass fiber filter followed by a polyurethane foam (PUF) plug some distance downstream in the direction of air flow. Typically, these are active samplers utilizing electric pumps to regulate the flow of sampling air to some predetermined rate, usually in the range of 300 - 400 cubic meters of air over a 24-hour sampling period. Unlike the stack sampling method, the particulate filter is not artificially heated, nor is there a condensing component where the sampled air is quickly cooled in order to force condensation of vapor phase semi-volatile organics.

Hence this sampling scheme should be more useful to the interpretation of the vapor phase/particle phase partitioning of PCDDs/Fs under ambient conditions. Such observations are made on the basis of the segregation of the dioxin congener in the filter versus the PUF plug. The PUF plug has been verified to efficiently trap vapor phase semi-volatile pesticides and organic species (Wagel, et al., 1989), and the glass fiber filter uses filter paper with porosities to 0.1 microns to collect particulate matter. Because sampling of PCDD/Fs is not instantaneous (i.e. real time measurement), but requires 24-hour air sampling to assure a level of detection of about 0.03 pg/m3, the interpretation of the V/P ratio should be construed as operationally defined rather than a direct empirical observation.

Under this assumption, the V/P ratio is relative and not absolute. The following is a review of ambient air sampling providing sufficient information of the relative V/P partitioning of PCDD/F congeners at ambient conditions. Table 3-4 provides a summary of the V/P ratio inferred from these reports.

Oehme, et al. (1986)
first described a method sensitive enough for the congener-specific measurement of PCDD/Fs at 0.1 pg/m3 levels of detection in ambient air. Such low levels of detection introduced the possibility of taking ambient air samples in the vicinity of known combustion sources of PCDD/Fs to reliably establish an association with sack emissions. Oehme tested the performance and reliability of an ambient air sampler consisting of a glass fiber filter followed by a polyurethane foam plug. Ambient air was sampled over a predetermined period after first spiking the filter with a known concentration of 13C12 labelled PCDD/F standards. This experiment was designed to determine the percent of the initial spiked labelled standard that could be recovered from the sampler after sampling 1000 m3 of ambient air.

The percent recovery of the standard was a measure of the collection and retention efficiency of the sampler. After collecting a sample, the particulate filter and the PUF plugs were extracted and analyzed separately. This was done in order to establish the particle phase and vapor phase partitioning of the PCDD/F congeners.

table Table 3-4. Percent distribution of dioxins and furans between vapor phase (V) and particulate phase (P) in ambient air as observed in ambient air sampling studies.
Oehme demonstrated that the sampling method was capable of a high degree of reliability in sampling sub-part per trillion concentrations of PCDD/Fs as indicated by highly satisfactory recovery of the isotopically labelled standards in the apparatus, e.g., 88 - 102% recoveries. From the results of separately analyzing the filter and the PUF, Oehme postulated on the typical distribution of PCDD/Fs between vapor and particles in ambient air. They suggested that TCDF and PeCDF were mainly present in the vapor phase, and HxCDD, HxCDF as well as the less volatile isomers of HpCDF, HpCDD, OCDF, and OCDD, were mainly present in the particle phase. Oehme took over 60 ambient air samples with this device in rural, suburban, and urban areas of Europe.
expand table Table V3 3-4

Eitzer and Hites (1989)
reported on the measurement of PCDD/Fs in the ambient atmosphere of Bloomington, Indiana while using a similarly configured ambient air sampling method, the General Metals Works PS-1 sampler. Ambient air is drawn through a glass fiber filter followed by a polyurethane foam plug (PUF). This was a long-term study designed to investigate the daily and seasonal variability of the compounds in the ambient air as measured at a single location, and to examine the vapor-phase, particulate-phase partitioning of the chlorinated congeners under ambient conditions. Samples were taken at four different sites over a 2-3 day sampling period until 1500 to 2400 m3 of ambient air volume had passed through the apparatus. Sampling was conducted monthly from August, 1985 through July, 1986. The quantitative method produced a limit of detection of the individual chlorinated congeners in the range of ~1 femtogram/m3. Eitzer and Hites (1989) operationally defined the vapor-phase/particle-bound phase of the chlorinated congeners as any compounds found in the PUF plug and the glass fiber filter, respectively. The V/P ratio was subject to certain restrictions of the sampling method, which the authors identified as: 1. Particles smaller than 0.1 microns would pass through the filter paper of the glass fiber particulate filter and be absorbed into the polyurothane foam; 2.

Diurnal temperature variation could cause particle-bound PCDD/Fs collected and retained in the filter to vaporize and be "blown-off" to the PUF plug by the passage of the sampled air stream; 3. At these relatively large sampling volumes of ambient air, it is possible that some breakthrough on the PUF plug occurs, and a portion of the PCDD/F sample is lost. The investigators were able to rule-out the latter condition through the addition of a XAD-2 resin trap after the PUF. This was one of the first reports on the congener-specific V/P partitioning in the ambient air under variable average ambient temperatures. Although they could find no seasonal effect on the total concentrations of PCDD/Fs, seasonal change in temperature did affect the V/P ratio. It was noted that during the warm summer months the V/P ratio was as great as 2:1, and during the cold winter months the V/P ratio could be <0.5. Thus, at warm temperatures most of the lower chlorinated congeners, e.g., mono through penta-chlorinated PCDD/Fs, were mostly found in the vapor phase and the hexa - octachlorinated congeners were mostly particulate-bound.

The colder winter temperatures produced the effect of more eventually splitting the V/P distribution of the lower chlorinated species. The higher chlorinated congeners, e.g., hexa-, hepta-, and octa-PCDD/Fs, mostly were found to be particle-bound at both the warm and cold temperatures. These quantitative results of the V/P ratio of individual congeners at three ambient air temperatures (3 C, 16 - 20 C, and >28 C) was again reported by Hites (1991), as shown in Table 3-4. Through these analyses, Eitzer and Hites (1989) and Hites (1991) found two dependant variables controlled the V/P ratio in ambient air: 1. the ambient air temperature; and 2. the vapor pressures of the PCDD/F congeners. The authors concluded that because the lower chlorinated compounds have higher vapor pressures, they will be found mostly in the vapor phase, and because the higher chlorinated congeners have lower vapor pressures, they will prevail in the ambient air bound to particulate matter.

Wagel, et al. (1989)
reported on the performance of the General Metals Works PS-1 sampler for the collection and retention of PCDD/Fs while sampling ambient air. This sampler configuration consists of a quartz glass fiber filter followed by a polyurethane foam (PUF) plug, and the investigators added an XAD-2 resin cartridge after the PUF. The addition of the XAD was a check on whether breakthrough of any PCDD/F congeners occurred from the PUF during sampling. The PS-1 is the sampler most often used in the U.S. to quantify PCDD/Fs in air under ambient conditions. The protocol of this research was to use two samplers co-located. The particulate filter of one sampler was spiked with 13C12-labelled PCDD/F congeners while the second sampler was used to provide background measurements of native (non-labelled) PCDD/Fs. Both units were operated to sample ambient air for 24-hours. The average ambient temperature during the sampling period was 24 C. Following the sampling the filter and PUF were removed and extracted according to published procedures (Wagel, et al., 1989). Performance of the PS-1 sampler was reported as percent recovery of the labelled standards initially spiked onto the particulate filter. The percent recovery was calculated by subtracting the background contributions from the total detected spike concentration and dividing by the concentration of the labelled standard initially added to the filter. The percent recoveries were reported in a range of from 85% to 124%, with an average recovery of 102%. This indicated a high degree of reliability in collecting and retaining PCDD/Fs in the sampler during the 24-hr sampling period.

A second series of experiments were conducted to investigate the distribution of PCDD/Fs within the sampling apparatus, e.g., the particulate filter versus the PUF plug, by extracting and analyzing the filter and PUF separately. Subject to the caveats previously discussed, the investigators made observations regarding the V/P ratio of PCDD/F congeners. It was observed that PCDD/Fs having 7-8 chlorines were mostly detected in the particulate filter, and lower chlorinated species were mostly detected in the PUF. Wagel, et al. (1989) suggested that it was possible that the lower chlorinated congeners volatilized from the particulate filter (somewhat affected by the rate of flow of the sampled air volume), and then were retained by the PUF. Furthermore, Wagel, et al. (1989) warned that if results of separately analyzing the filter and PUF are used to derive a vapor phase and particle phase partitioning of the PCDD/Fs under ambient conditions, then this may give erroneously high estimates of the amount present in vapor phase.

Harless and Lewis (1992)
have quantitatively evaluated the performance of the General Metals Works PS-1 sampler for the trace-level measurement of PCDD/Fs in ambient air, adding to the growing evidence that results are actual measurements and not an artifact of the sampling method. In this study, three samplers were used in the same general vicinity, and were operated for a 24-hour period until an air volume of 350 - 400 m3 had passed through the system. The quartz glass fiber particulate filters of two of the samplers were then spiked with 13 C12 labeled PCDD/F congener with a known concentration after the 24-hour sampling period. The three samplers were then operated another 24-hours. The samplers were then shut down, and the filters and PUF plugs were removed and extracted and analyzed for PCDD/Fs separately according to prescribed procedures. A separate series of experiments involved precleaning the glass fiber particulate filters, and adding the isotopically labelled PCDD/F spike to the filter prior to sampling for seven days until about 2660 m3 of ambient air had been sampled.

Results of this study confirmed the accuracy and reliability of the PS-1 sampler for collecting and retaining PCDD/Fs at sub-ppt concentrations in ambient air. Performance was defined as the percent of the initial concentration of the labelled isotope recovered in the sampling apparatus following the operation over the predetermined sampling period. The average efficiency of recovery of the 0.8 ng 13C12-1,2,3,4,-TCDD isotope that was spiked onto the filter prior to sampling was 91%, and similar efficiencies were observed for the recovery of the other labeled PCDD/Fs. Additionally, Harless and Lewis (1992) used the spiking system to observe the distribution of PCDD/Fs in the filter and the PUF after sampling 400 m3 of ambient air. It was observed that most of the hepta-, and octa-CDD/Fs were retained by the glass fiber filter, indicating that these compounds were primarily particulate-bound, and most of the tetra-, penta-, and hexa-CDD/Fs volatilized and were collected by the PUF plug. When partitioning was observed on a congener-specific basis, significant differences were observed in the V/P ratio, as shown in Table 3-4.

Hunt and Maisel (1990)
reported on the ambient air measurement of PCDD/Fs in a northeastern U.S. urban coastal environment during the fall and winter seasons. Isomer-specific sampling was conducted with the General Metal Works PS-1 sampler in and around Bridgeport, Connecticut from November, 1987 through January, 1988. Nine sampling sessions consisting of a total of 43 ambient air samples were taken in this study. Each sampling session was conducted either over a 24-hour or 72-hour period until about 350 m3 and 600 m3 of air volume had passed through the sampler. Hunt and Maisel (1990) reported on the typical vapor phase/particle bound partitioning of individual congeners during cold ambient air temperatures. The V/P ratio was based on the results of separately analyzing the PUF plugs and the glass fiber particulate filters for the presence of PCDD/Fs. From these data, the investigators concluded that greater than 92% of all the congeners of PCDD/Fs were particulate bound (operationally defined as detected in the particulate filter). The 2,3,7,8-TCDD isomer was not detected in any of the 43 collected samples (reported limit of detection was 5-20 fg/m3). The particulate bound distribution (reported as a percent of the detected concentration) for some of the other congeners were as follows:

Diagram V3 3-1

The vapor phase/ particle bound distribution observed in this study is probably controlled by the cold January temperatures from which these observations were derived (average temperature = -5 C).

At a later date, Hunt and Maisel (1992) conducted ambient air monitoring of PCDD/Fs in multiple locations in the warm climate of southern California for the State of California Air Resources Board (CARB). Ambient air samplers, e.g., the General Metal Works PS-1 sampler, were primarily placed in areas of high population density that contained known combustion sources of PCDD/Fs, but sites were also sampled that were considered removed from the influences of any local sources. The purpose of the study was to evaluate the congener-specific spacial distribution of PCDD/Fs in ambient air near environmental sources of the compounds, and in remote locations, in order to provide a baseline to evaluate population exposures within the region. Monitoring sites were established at eight locations in the South Coast Air Basin in and around the city of Los Angeles.

Nine discrete sample sets were collected from December, 1987 through March, 1989. The authors defined a sample set as consisting of five to seven stations at which one or two co-located samplers were operated. Microscale meteorological data was collected during sampling to include wind speed, wind direction, barometric pressure, and temperature. One sampling site was chosen to investigate the distribution of PCDD/Fs in ambient air where average ambient temperatures ranged from 16-20 C. This was done by the usual procedure of separately analyzing the filter and the PUF and making the assumption that what is detected in the glass fiber filter is particulate bound, and what is trapped in the PUF is in vapor phase. The authors noted that under these conditions, the V/P partitioning is operationally defined by the ambient air sampling system, and therefore may not be a true indication of the partitioning in the atmosphere.

The majority of the hexa through octa CDD/F congeners were detected in the filter, and the authors observed that they were mainly associated with particulate matter. The authors found these observations were consistent with the V/P ratio observed by Eitzer and Hites (1989) in warm climate conditions. In addition, the authors noted that these observations give further evidence that vapor pressures of the specific PCDD/F compounds and ambient air temperatures strongly influence the V/P partitioning. Therefore the tetra- and penta-CDD/Fs are expected to predominate in vapor phase during warm seasons. However, during the cold temperatures of the winter season these congeners are expected to be primarily associated with particulate matter in the ambient air.

Bobet, et. al. (1990)
reported the results of an ambient air monitoring network operated by Environment Canada to temporally measure PCDD/Fs in the ambient air in southwestern Ontario, Canada. The intent of the study was to monitor possible environmental impacts of a large refuse-derived fuel municipal waste combustor operational in the City of Detroit, Michigan. The ambient air monitoring network consisted on two stations, one in Windsor, Ontario, and the other located in the Walpol Island Indian Reservation 18 km to the northeast of Windsor.

The former site was considered in an urban area near the expected point of maximum impact from the stack emissions from the MWC, and the other site was considered rural, and away from the influence of any stationary combustion source. PCDD/F samples were collected once every 24 days using an high-volume ambient air sampler consisting of a Teflon-coated glass fiber particulate filter and a PUF adsorbent trap. Ambient air was sampled over a 24-hour period from July, 1987 to August, 1988 with a total sample volume of 800 - 1000 m3 of air. From August on, the samplers were operated over a 48-hour period, and 1600 - 2000 m3 of air passed through the sampler.

Mean total concentrations of PCDD/Fs were compared between the urban and rural sites, and Bobet observed that concentrations measured at the urban site were 4 -20 times greater than at the rural site. Additionally, the V/P partitioning of PCDD/Fs (as operationally defined by detection in the PUF verses detection in the filter) was investigated at both sampling stations. Bobet stated that the V/P may be influenced by "blow-off" of particulate from the filter to the PUF, and/or the passage of particulate matter <0.1 microns from the filter to the PUF, and if this is the case, then the vapor phase partitioning may be too great as interpreted by the method. Under these circumstances the authors suggested that the V/P partitioning should be considered as roughly representative of the vapor/particulate phases in the ambient air. On a total concentration basis, and on a total of 12 separate ambient air samples, the investigators found the following average percent vapor phase versus percent particle phase partitioning of the PCDD/F homologues at the Windsor, Ontario station:

Diagram V3 3-2

At the rural Walpole Island station, no TCDD, PeCDD or TCDF - OCDF were detected in any of the 5 separate ambient air samples. All of the detected HxCDD, HpCDD and OCDD was found in the particulate filter indicating a V/P distribution of 0% V/ 100% P for these compounds. The authors did not report the average ambient air temperature at the two stations.

3.2.4.4. Discussion of the vapor/particle partitioning in ambient air sampling studies

The studies that have been reviewed here indicate the following:

The high-volume ambient air sampler consisting of a glass fiber particulate filter and polyurethane foam absorbent trap is a reliable method for the collection and retention of PCDDs/Fs in ambient air.

Current analytical methods assure detection limits, on a congener specific basis, of about 0.03 pg/m3.

Experiments involving the recovery of isotopically labelled PCDD/Fs within the sampler after 24-hours operation indicate that the sampler does not create artifacts representative of either sample losses or the synthesis of dioxin.

Because the sampler is not artificially heated or cooled, but is allowed to operate at existing ambient air temperatures during sampling sessions, the method can be used to imply the vapor phase and particle bound partitioning of PCDD/Fs in ambient air. This is accomplished by separately extracting and analyzing the glass fiber filter and the polyurethane foam for the presence of PCDD/F congeners.

However, the V/P ratio interpreted from these results is operationally defined. This will only give an approximate indication of the V/P ratio since mass transfer between the particulate filter and the vapor trap cannot be ruled out. The particulate filter paper porosity is 3 0.1 microns, and therefore it is possible that aerosol particles with diameters < 0.1 microns will pass through the filter and be trapped in the polyurethane foam plug. If this is the case, then the percent observed in vapor phase will be overestimated.

The method involves ambient air sampling at a relatively high sample volume, around 300-400 m3 of air, over a 24-hour period. It is possible that PCDD/Fs that are not sorbed to particulate matter captured in the filter may be volatilized by subtle changes in ambient temperature, and that PCDD/F in the vapor phase may be carried with the sampling air flow to the PUF sorbent trap. If this were to occur, then the interpretation of the percent of the PCDD/Fs partitioned to the vapor phase would be an over estimate. Unfortunately there are no empirical data that have demonstrated that any of these effects may actually occur.

3.2.4.5. Theoretical prediction of vapor/particle partitioning of PCDD/Fs under ambient conditions

Bidleman (1988)
offers a theoretical construct for estimating the vapor phase/particle bound partitioning of PCDD/Fs in ambient air. Bidleman presents the theory that a portion of the semivolatile compounds found in ambient air are freely exchangeable between the vapor and particle phases. Bidleman defines a second portion, the nonexchangeable fraction, as the quantity that is strongly and irreversibly adsorbed to particulate matter, and is not at equilibrium with a corresponding vapor phase. Bidleman cites an earlier model by Junge (1977), a theoretical model based on adsorption theory, which mathematically described the exchangeable fraction of the semivolatile organic compound adsorbed to aerosol particles as a function of solute saturation vapor pressure and total surface area of atmospheric aerosol particles available for adsorption.

This is given by:

Equation V3 3-2

Although Junge treated the term 'c' in Equation (3-2) as a constant, e.g., c=1.7 E-4 atm-cm, Bidleman notes that it actually is variable and quite dependent on the chemical's sorbate molecular weight, the surface concentration of the chemical on aerosol particles (assuming monolayer coverage), and the difference between the heat of desorption from the surface of a particle and the heat of vaporization of the liquid-phase sorbate.

Bidleman (1988) poses the question as to whether it is the chemical's sub-cooled liquid vapor pressure (Pl) or the chemical's crystalline solid vapor pressure (Ps) that ultimately controls the rate of adsorption to aerosol particles. Pl and Ps are related according to Equation (3-3) developed by Bidleman (1988). The sub-cooled liquid vapor pressure is estimated by extrapolating below the melting point of the compound.

<empty>Equation V3 3-3

Bidleman notes that a satisfactory estimate of _Sf/R observed in other treatments of this subject is 6.79. This can be substituted for _Sf/R in Equation (3-3), and used as a constant. Bidleman argues that the use of Pl in Junge's equation, the sub-cooled liquid vapor pressure, makes the most accurate estimation of the vapor phase/particle bound partitioning of semi-volatile organic compounds in ambient air.

To support his argument, Bidleman gives the example of the comparison of the application of the crystalline solid (Ps) versus the sub-cooled liquid (Pl) vapor pressure of TCDD in Junge's equation to estimate the V/P partitioning at an ambient air temperature of 20 C in an urban air shed. The Ps predicts a V/P partitioning of 0%/100%, whereas the Pl predicts a V/P of 20%/80%.

Bidleman then compares these predictions against the V/P partitioning of TCDD as observed by Eitzer and Hites (1986) from the sampling of ambient air for PCDD/Fs in Bloomington, Indiana.

His conclusions are that the prediction of the V/P ratio using the sub-cooled liquid vapor pressure of TCDD best fits the observed partitioning interpreted from directly measuring PCDD/Fs in ambient air, e.g., most tetra-, and penta-CDD congeners are prevalent in the vapor phase, and the higher chlorinated congeners are mainly particle bound (as operationally defined by the sampling method).

Calculations of f (the fraction that is bound to particulate) from Equation (3-2) can be made on a congener-specific basis for the PCDD/Fs. The estimate of 1.7 E-4 atm-cm for the value c can be assumed from the work of Junge (1977) as cited by Bidleman (1988). The sub-cooled vapor pressures can be converted from the crystalline solid vapor pressures of the specific congeners found in Volume 2, Chapter 2, of this assessment, by applying Equation (3-3).

The melting points of the specific congeners are also referenced in Chapter 2, Volume 2. Bidleman (1988) provides estimates of average total surface areas of aerosol particles relative to average total volume of air (cm2/cm3), the term ST in Equation (3-2), citing a study by Whitby (1978).

In addition, Whitby estimated the average total volume of aerosol particles per volume of air (Vt = cm3 particles/cm3 of air). Whitby's (1978) calculations varied according to the density of aerosol particles in the ambient air in different air sheds.

These were as follows (units of ST of cm2/cm3, VT of cm3/cm3):

Diagram V3 3-3

Bidleman noted that if the average particle density of aerosol particles suspended in urban air is assumed to be 1.4 grams/cm3, then the surface area of the average urban aerosol particles is 11 m2/g, and the average total suspended particulate is 98 g/m3, following the calculations of Whitby. This was regarded as being in close agreement to the average monitored total suspended particulate of 79 g/m3 for 46 cities surveyed in the United States in 1976.

Therefore, Whitby's values for ST were judged by Bidleman to be adequate for purposes of calculating and estimating f . If these assumptions are applied to the variables in Equation (3-2), the value for f , the fraction of the PCDD/F congener reversibly adsorbed to particles in ambient air, can be calculated. Table 3-5 shows the calculated fraction of the congener that is bound to particles suspended in ambient air over a range of air shed classifications.

table Table 3-5 Fractions of dioxins and furans calculated to partition to particles in various classifications of ambient air using the method of Bidleman (1988), Junge (1977), and Whitby (1978.
The values of the fraction of PCDD/Fs adsorbed to particulate in the ambient air can be compared against measurements of partitioning of the PCDD/Fs in ambient air as a simple means of validating these calculations. Bidleman (1988) compared theoretical predictions of the fraction of TCDD partitioning to suspended particulate in urban air to the measurements taken by Eitzer and Hites (1986) in an urban setting as a way of qualitatively addressing the "reasonableness" of his calculations. He found that the model's estimation of the fraction of PCDD/Fs bound to particles (based primarily on adsorption theory and using the sub-cooled liquid vapor pressure of 2,3,7,8-TCDD) did agree with the ambient measurements of Eitzer and Hites (1986) at 20 C.
expand table Table V3 3-5
Therefore the model was judged to be reasonable in it's estimation of the physical state partitioning of PCDD/Fs in ambient air.

3.2.4.6. Discussion of vapor/particle partitioning

This subsection has reviewed stack testing data, ambient air sampling data, and theory rooted in basic physical chemistry that either imply, directly deduce or theoretically calculate the V/P partitioning in the ambient air. From this review it is generally concluded that:

1. The stack test methods in use today to monitor and measure the concentration of PCDD/Fs emitted to the air from combustion sources do not provide a credible basis for assuming the vapor phase and particle bound partitioning at the point of release. There is no consistent pattern to the interpretation of V/P based on where the PCDD/F segregates in the instrument, e.g., the glass fiber filter or the XAD resin. Factors that may contribute to this are: the relatively long residence time spent traversing the stack interior; the probe to the instrument is inserted into a relatively hostile environment of the hot combustion gas; the static temperature of the particulate filter caused by heating the particulate filter housing; the fact that located between the particulate trap and the vapor trap is a condensing section consisting of glass tubing surrounded by an ice bath.

2. On the other hand, the ambient air sampling methods do give an approximate indication of the V/P ratio that seems to be responsive to changes in temperature, and degree of chlorination of the PCDD/Fs. This is in accordance with what would be expected from their individual vapor pressures. There is no artificial heating or cooling of any component of the sampler. The sampler is exposed to actual temperature, pressure, and humidity of the ambient air.

This removes the possibility that the vapor phase-particle bound partitioning, operationally defined as the compound segregating to the particulate trap and vapor trap, is actually an artifact induced by artificial heating and cooling within the system. Therefore the methods present a realistic picture of partitioning under variable ambient conditions. However, the method has certain limitations that currently prevent deriving a true measurement of V/P partitioning in the ambient air. Among these limitations are:

a. The glass fiber filter designed to capture and retain particulate matter has filter pours down to 0.1 m diameter. Particles less than this diameter will pass through the filter and be retained in the polyurethane foam vapor trap downstream. If this is the case, the amount of PCDD/Fs observed to be particle bound would be underestimated, and the amount observed to be in vapor phase would be overestimated.

b. The relatively high sampling volume passed through the system (200 to 400 m3 of air per 24 hours) may redistribute the more volatile congeners from the filter to the absorbent trap by a process known as 'blow-off'.

3. Until sampling methods are improved and modified such that they give results that indicate the true V/P ratio of PCDD/Fs in ambient air, the theoretical construct described by Bidleman (1988; and detailed above) is used to calculate the V/P ratio for purposes of air dispersion and deposition modeling of emissions from the hypothetical case demonstrated in Chapter 5.

Key advantages to the theoretical approach are that the theoretical construct relies on current adsorption theory, considers the molecular weight and the degree of halogenation of the congeners, uses the boiling points and vapor pressures of the congeners, and uses the availability of surface area for adsorption of atmospheric particles that correspond to a variety of ambient air shed classifications having variable particulate matter densities.

3.2.5. Estimation of the Concentration of Dioxin-Like Compounds in Incineration Ash

The ash that is collected by the particulate matter control device preceding the stack is known conventionally as fly ash. Fly ash is the airborne combustion residue from burning the fuel. Bottom ash is the ash residue that results from the combustion of the organic solids within the combustion chamber, and usually is collected below a grate system used to convey combustible fuels into the fire zone, or is collected at the bottom of the combustion chamber.

In general, there are many factors that may influence the formation of particulate matter known as fly ash from the incineration of organic wastes. Among these factors are: the heating value of the incinerated material (BTU/kg), the percent moisture in the fuel, the furnace temperature and combustion efficiency, and the efficiency of particulate matter capture by the air pollution control device (Brunner, 1984; OTA, 1989).

Fly ash, and not the bottom ash, contains most, if not all, the dioxin-like congeners. This can be explained by the synthesis of dioxin that occurs on the reactive surface of fly ash. Therefore, the following estimation of the ash generation rate, and the concentration of dioxin-like compounds in the ash particles, will focus solely on fly ash to the exclusion of bottom ash. Because bottom ash is mostly free of these contaminants, and is about 10 to 100 times the mass of fly ash, the mixing of fly ash with bottom ash will dilute the concentration of dioxin by about a factor of 10 - 100.

Estimation of the mass of fly ash generated, and concentration of dioxin-like compounds can be determined by the following (if no actual data exists):

1. Determine the mass of fly ash generated per day at the facility. This can be estimated from the percent control of particulate matter (PM) of the air pollution control device (APCD) installed at the facility. For example, if a combustor emits 0.5 kg of particulate matter per hour of operation, then 12 kg of PM is released from the stack in one day. If PM is controlled by 99%, then this rate of emission represents one percent of the fly ash generated by the combustion process. The amount of fly ash that is collected by the APCD would be 100 times the amount emitted, or 1200 kg/day.

2. Estimate the congener-specific concentration of PCDD/Fs contained in the collected fly ash. This is done by assuming that what is prevented from exiting the stack is contained in the fly ash collected by the pollution control device. If, for example, 10 picograms PCDD/F is emitted per gram of PM from the facility per day, and the APCD reduces emissions by 99%, then 100 times more PCDD/F concentration, or 1000 picograms PCDD/F per gram fly ash, would be in the collected fly ash. If the concentration of dioxin in emitted fly ash and the percent control of dioxin are known, then the concentration of dioxin in the mass of collected fly ash can be estimated. It is important to make such estimations in order to evaluate the potential environmental impact of ash management practices before the operation of the facility, and to select appropriate disposal practices to preclude future adverse conditions from arising.

3. Now estimate total mass, including fly and bottom ash, and final concentrations. If bottom ash mass is estimated at ten times fly ash, than the total ash generated in this example would be 1200 + 1200*10 = 13,200 kg/day. If fly and bottom ash were mixed for disposal, which is common, than the average concentration of the total ash would be one-tenth that estimated for fly ash.

The hypothetical example in Chapter 5 does not assess impacts associated with ash disposal. Section 4.4.3.1 of Chapter 4 describes procedures for estimating impacts from ash disposal given ash concentrations and mass generated.

3.3. AIR DISPERSION/DEPOSITION MODELING OF THE STACK GAS EMISSIONS OF DIOXIN-LIKE COMPOUNDS

It has been customary for EPA to use air dispersion/deposition models to estimate the atmospheric transport, the deposition flux, and the ambient air concentrations of specific pollutants attributable to smokestack emissions from an industrial combustion source. Air dispersion models are mathematical constructs that approximate the physical and chemical processes occurring in the atmosphere that directly influence the dispersion of gaseous and particulate emissions from smokestacks of stationary combustion sources.

These models are computer programs encompassing a series of partial differential and algebraic equations to calculate the dispersion and deposition of the emissions. Concentration and deposition isopleths of the pollutants discharged from the stack are computed at specified distances from the smokestack. These quantities are used to estimate the magnitude of potential exposures to the human receptor.

Numerous dispersion/deposition models have been developed. This document focuses on the COMPDEP model. COMPDEP was first described in EPA (1990b). Recent revisions to the computer code were made, and this newer version was used to generate the results for the hypothetical incinerator of this assessment. A principal change that was made allows the user to define 10 particle size categories. Earlier versions allowed only 3 size categories. Use of COMPDEP is this assessment is not intended to imply that COMPDEP is the only acceptable model to use in the analysis of ambient air concentrations, and wet and dry deposition.

Subsection 3.3.1 below presents an overview of the dispersion and deposition algorithms in the COMPDEP model. Subsection 3.3.2 discusses dry deposition fluxes, including pertinent assumptions made in the application of the COMPDEP model for the hypothetical combustor demonstrated in Chapter 5. Subsection 3.3.3 discusses particle size distributions for emitted particles. Subsection 3.3.4 discusses wet deposition, again noting key assumptions for the hypothetical combustor. Subsection 3.3.5. closes the section with guidance indicating that the COMPDEP model should be run twice for assessments - once to model particle fate and transport, and once for the vapor phase.

3.3.1. Basic Physical Principles Used to Estimate Atmospheric Dispersion/Deposition of Stack Emissions

Air dispersion/deposition models use the basic physical processes of advection, turbulent diffusion, and removal to estimate the atmospheric transport, resulting ambient air concentration, and settling of particles. Advection describes the physical movement of the air pollutants by the horizontal movement of wind. Turbulent diffusion is the "spreading" of the emissions plume with distance from the stack due to multi-directional fluctuations in air movement. Removal refers to mechanisms which remove emissions from the atmosphere. This can be caused by the force of gravity exerted on the particle mass, Brownian movement of aerosol particles, and scavenging of particles. Scavenging is the removal of particles or vapors by precipitation.

COMPDEP contains modifications of the Industrial Source Complex model (Short-Term version), and COMPLEX I to incorporate algorithms to estimate dispersion, and resulting ambient air concentrations and wet and dry deposition flux. COMPLEX I is a second level screening model applicable to stationary combustion sources located in complex and rolling topography (EPA, 1986a).

The COMPDEP model was developed by EPA to provide estimates of air concentrations and deposition rates of the stack emissions of contaminants from industrial sources located in varied terrain (e.g., from simple to complex terrain). Simple and complex terrain are defined as topogragraphy that is either below or above the effective stack height of the source (Turner, 1986).

To account for pollutant deposition, the concentration algorithms in COMPLEX 1 were replaced with those from the Multiple Point Source Algorithm with Terrain Adjustments Including Deposition and Sedimentation (MPTER-DS) model (Rao and Sutterfield, 1982). The MPTER-DS algorithms incorporate the gradient transfer theory described by Rao (1981), and are extensions of the traditional Gaussian plume algorithms.

The dispersion algorithms contained in the Industrial Source Complex, Short-term version (ISCST), have been incorporated in COMPDEP to analyze ground-level receptors located below the height of the emission plume (EPA, 1986b). COMPDEP uses the generalized Briggs (1975, 1979) equation to estimate plume-rise and downwind dispersion as a function of wind speed and atmospheric stability.

A wind-profile exponent law is used to adjust the observed mean wind speed from the measurement height to the emission height for the plume rise and pollutant concentration calculations. The Pasquill-Gifford curves are used to calculate lateral and vertical plume spread (EPA, 1986a). These curves are based on Pasquill's definitions of atmospheric stability classes, e.g., extremely unstable, moderately unstable, slightly unstable, neutral, slightly stable, and moderately stable, that correspond to various intensities of solar radiation and wind speeds (Seinfeld, 1986). The incorporation of these two basic models into COMPDEP permits analysis of a source located in all types of terrain.

3.3.2. Estimation of Dry Surface Deposition Flux

Dry deposition is one removal process that is simulated by the COMPDEP model. Dry deposition refers to the transfer of airborne particulate matter to the earth's surface (including water, soil, and vegetation) whereby it is removed from the atmosphere. The deposition of vapor-phase contaminants is not considered in the COMPDEP model.

The general processes controlling the transfer of particulate from some height above the surface through the surface layer down to the immediate vicinity of the surface are the forces of gravity and turbulent diffusion (Seinfeld, 1986), followed by diffusion through the laminar sub-layer (defined as a thickness of 10-1 to 10-2 cm) to the surface.

The rate at which contaminants sorbed to atmospheric particulates are removed by the physical forces of gravity, atmospheric turbulence, and diffusion is termed the "deposition flux" (Kapahi, 1991), and is mathematically represented by Fd. The deposition flux, Fd, is a function of the concentration of the chemical contaminant on particulate, Co, and the settling velocity of the contaminated particles can be defined by:

Equation V3 3-4

In general, Chamberlain and Chadwick (1953) first defined the settling velocity, Vd, as the quotient of the deposition flux, Fd, divided by the airborne concentration, Co:

Equation V3 3-5

Sehmel (1980)
noted that the value for Fd in Equation (3-5) has a minus sign because the downward flux is negative, whereas the deposition velocity is positive. By this relationship, Chamberlain and Chadwick (1953) first introduced the concept of plume depletion: as the plume emission is dispersed with downwind distance from the stack, the deposition flux decreases with distance from the source.

The basic dynamics in the physics of modeling dry deposition have not changed significantly since Sehmel's (1980) comprehensive scientific review. The factors that most influence the predicted deposition flux can be divided as being either meteorological influences, or the influences of the properties of the pollutant under analysis. Meteorological influences include the friction velocity, represented as o, and the aerodynamic surface roughness, represented as zo.

These terms are used to describe the wind speed profile above the Earths surface. In most cases, the analyst uses a graphical procedure to determine values for o and zo. If the logarithm of wind speed is plotted for near neutral atmospheric stability as a function of height from the surface, then the values for the constant zo is fitted to a straight line on a semi-logarithmic scale. This can be described mathematically by Equation (3-6). In most cases, the friction velocity is a percentage of the wind speed.

Equation V3 3-6

As a general rule, particles greater than 30 micrometers (m) in diameter will be removed from the atmosphere primarily by the force of gravity, whereas particles less than 30 m will be removed primarily by atmospheric turbulence. The deposition flux for the smaller particles is influenced by many factors, including: the distribution of particles by diameter and density; assumptions of atmospheric turbulence; the friction of the ground surface and the height of the stack release of emissions. Deposition flux is also affected by the partitioning properties of the pollutant. These properties will determine how much of the pollutant is sorbed to the particle and how much is in the vapor phase. A detailed list of the many factors that can affect dry deposition is shown in Table 3-6. The COMPDEP estimates dry deposition flux based on empirical associations developed by Sehmel (1980) and Sehmel and Hodgson (1978) relating the deposition flux to the deposition velocity of particles. The downward motion represented by deposition velocity is controlled by the gravitational settling velocity, atmospheric resistance, surface resistance and the atmospheric surface friction layer. This model assumes that a fraction of the particulate comes into contact with the ground surface by the combined processes of gravitational settling, atmospheric turbulence, and Brownian diffusion. The COMPDEP model contains enhancements to calculate dry deposition flux using a computerized routine developed by the State of California Air Resources Board (CARB, 1986).

table Table 3-6 Factors that influence the dry deposition removal rate in the atmosphere.
The CARB algorithms represent Sehmel's (1980) empirical relationships for transfer resistances as a function of particle size, density, surface roughness, and friction velocity.

In the CARB model, integrated resistances to mass transfer are computed within two layers. In the first layer, which extends from one centimeter to one meter above the surface, atmospheric turbulence dominates mass transfer. In the second layer, which lies within one centimeter of the surface, the resistance to mass transfer is derived from particle deposition measurements that were taken in a wind tunnel over various surfaces using mono-dispersed particles (Sehmel, 1980; Sehmel and Hodgson (1978).
expand table Table V3 3-6

Despite what is currently known about the physical and chemical processes that influence the final deposition flux of particles released from a stationary combustion source, a more thorough understanding of the influence of particle size on deposition velocity is needed. In Sehmel's (1980) review of settling velocities corresponding to particle diameter it was noted that the range of values spanned several orders of magnitude.

This complicates efforts to make generalizations of Vd by particle diameter for air modelling purposes. Although dry particle deposition velocities have been estimated from both field studies and laboratory experiments, derived velocities are limited and highly uncertain. This is due largely to the complex and variable array of factors that can influence the rate of deposition (as depicted in Table 3-6).

In the general classification of particles, particles < 2.5 micrometers (m m) in diameter are referred to a "fine particles", and those > 2.5 m m are "coarse particles". Sehmel (1980) offers the most current review of dry deposition settling velocities for a variety of depositing materials having a broad range of particle diameters. This summary appears in Table 3-7.

For the example application of the COMPDEP model in Chapter 5, particles less than 2 m m were represented by a 1 m m size and were calculated by COMPDEP to deposit at a velocity of about 0.007 cm/sec. Particles between 2 and 10 m m were represented by a 6.78 m m size and were calculated to deposit at a velocity of about 0.3 cm/sec. Finally, particles greater than 10 m m were represented by a 20 m m size and were calculated to deposit at a velocity of 2.5 cm/sec, although the variable ambient air temperature resulted in more variable calculations. The derivation of these particle size representations is given in the next section.

3.3.3. Estimation of the Particle Size Distribution in the Stack Emissions

table Table 3-7 A summary of dry deposition velocities for particles.
Certain inferences must be made concerning the distribution of particulate differentiated on the basis of particle diameter before the COMPDEP program can predict deposition flux of the dioxin-like congeners.

The diameters of small particles comprising particulate matter in stack emissions are usually measured in units of one millionth of a meter (micrometer, commonly called micron, abbreviated by the letters m.

Unfortunately,
few studies describe the distribution of particulate matter entrained in the emissions from various combustion technologies broken down and fractionated by particle diameter.
expand table Table V3 3-7

The distribution of particulate matter by particle diameter will differ from one combustion process to another, and is greatly dependent on such factors as:

1) the efficiency of various air pollution control devices,

2) the composition of the feed/fuel,

3) the design of the combustion chamber,

4) the amount of air used to sustain combustion, and

5) the temperature of combustion.

Table 3-8 gives an example of a particle diameter distribution as measured at a stack on an incinerator. This example distribution will be assumed for the hypothetical incinerator. Although the COMPDEP model can simulate up to 10 particle size categories, only three particle sizes are assumed for the model runs of the demonstration in this assessment. These three sizes are generalized from the data in Table 3-8:

Diagram V3 3-4

After selecting the particle size distribution, it is necessary to calculate the mass emission rate of the particulate-bound congeners of PCDDs/Fs by particle size category. This is accomplished by calculating the proportion of surface area (available for adsorption of PCDD/Fs) for a given particle diameter.

table Table 3-8 Typical particle size distribution in particulate emissions from incineration.  
expand table Table V3 3-8

The ratio of the surface area to volume is proportional to the ratio of the surface area to weight for a particle with a given radius. Multiplying this proportion times the weight fraction of particles of a specific diameter (m) gives a value that approximates the amount of surface area available for chemical adsorption. The surface area to volume ratio can be described as follows:

Diagram V3 3-5

Dividing the surface area for each particle category by the total available surface area for all particles gives an estimation of the fraction of total area on any size particle. Multiplication of the emission rate of the dioxin-like congener times the fraction of available surface area will estimate the emission rate of the pollutant per particle size. The fraction of total surface area was computed for the three particle size categories. The fraction of total surface areas for the ranges of particle diameters are summed with each particle size category to represent a single fraction of total surface area for the given particle size category, as follows:

Diagram V3 3-6

Thus by these assumptions, 87.5% of the emission rate of the dioxin-like congener is calculated to be associated with particles less than L 2 m in diameter, 9.5% of the emission is associated with the particle size of > 2 to L 10 m, and only 3% of the emission is associated with particles greater than 10 m. To assist in deposition modeling of the emissions from the hypothetical incinerator, the particle size distribution is further simplified by assuming a median particle diameter to represent each broad particle size category, as follows:

Diagram V3 3-7

3.3.4. Estimation of Wet Deposition Flux

Wet deposition occurs by precipitation (rain, hail, snow) physically washing out the chemically contaminated particulate and vapors from the atmosphere. Vapor scavenging is not yet well understood and is not addressed in the COMPDEP model. The remainder of this discussion refers only to the wet deposition of particles.

Wet deposition flux depends primarily on the fraction of the time precipitation occurs and the fraction of material removed by precipitation per unit of time by particle size. Based on these relationships, scavenging coefficients were developed by Cramer (1986) for varying types and intensities of precipitation relative to different particle diameters by incorporating the observations of Radke, et al. (1980) in a study of scavenging of aerosol particles by precipitation. The principal assumptions made in computing wet deposition flux are:

(1) The intensity of precipitation is constant over the entire path between the source and the receptor;

(2) The precipitation originates at a level above the top of the emission plume so that the precipitation passes vertically through the entire plume;

(3) The flux is computed on the bases of fraction of the hour precipitation occurs as determined by hourly precipitation measurements compiled by the National Weather Service.

The remaining fraction (1-f) is subject only to dry deposition processes. Thus no dry deposition occurs during hours of steady precipitation, and dry deposition occurs between the periods of precipitation. Wet deposition flux is estimated using Equation (3-7), in which case the total weight of material in settling category n results from the washout by rain at a distance x downwind from the stack release:

Equation V3 3-7

The relationship between the scavenging coefficient, L (n,j), and the particle size and precipitation intensity was derived from the review of wet deposition studies of aerosol particles by Cramer (1986). Table 3-9 displays the scavenging coefficients assigned to the generalized particle size categories and precipitation events used for computing estimates of wet deposition in the application of the COMPDEP for the demonstration scenarios in Chapter 5.

The value of f in Equation (3-7) is applied to each hour of reported precipitation as determined from the National Weather Service meteorological observations. In computing the dry deposition which occurs between the periods of precipitation, a factor of (1-f) was used to estimate the fraction of the material that is subject to dry deposition. The same scavenging coefficient was used for both rainfall and snowfall.