Solid Phases of

Sep 7, 2005 - In addition, the removal efficiencies of vapor- and solid- phase PCDD/Fs are 84.5% and-13.4% in IWI-2, respectively. The results also in...
0 downloads 0 Views 815KB Size
Download PDF
Environ. Sci. Technol. 2005, 39, 8023-8031

Evaluation of PCDD/F Congener Partition in Vapor/Solid Phases of Waste Incinerator Flue Gases KAI HSIEN CHI AND MOO BEEN CHANG* Graduate Institute of Environmental Engineering, National Central University, Chungli 320, Taiwan

Activated carbon injection (ACI) is commonly used to control PCDD/F (polychlorinated dibenzo-p-dioxins and polychlorinated dibenzofurans) emissions from stationary sources. In this study, the characteristics of PCDD/Fs emitted from one municipal waste incinerator (MWI) and two industrial waste incinerators (IWI-1 and IWI-2) that apply activated carbon systems for controlling the emissions are investigated via intensive stack sampling. MWI and IWI-1 are equipped with ACI and bag filters (BF) while IWI-2 is equipped with a fixed activated carbon bed (FCB). Results indicate that most PCDD/Fs in flue gas downstream of ACI+BF exist in vapor phase (over 90%) while most PCDD/ Fs exist in solid phase (over 60%) downstream of FCB. For MWI and IWI-1, the removal efficiencies of vapor and solidphase PCDD/Fs are 98.5-99.6% and 99.8-99.9%, respectively. In addition, the removal efficiencies of vapor- and solidphase PCDD/Fs are 84.5% and -13.4% in IWI-2, respectively. The results also indicate that the partition of vapor/solidphase PCDD/F is affected by the type of the air pollutant control devices (APCDs) applied upstream and the particulate matter concentration in flue gas. On the basis of the sampling results of waste incinerators, this study preliminarily establishes the equations for predicting vapor/solidphase PCDD/F partition in flue gases downstream of various APCDs including cyclone (CY), electrostatic precipitator (EP), FCB, ACI+BF, and selective catalytic reduction system (SCR).

Introduction PCDD (polychlorinated dibenzo-p-dioxins) and PCDF (polychlorinated dibenzofurans) are commonly known as dioxin which has been listed as a persistent organic pollutants (POPs). The seventeen PCDD/F congeners with chlorine substitution in 2,3,7,8 positions are most toxic to humans. The vapor pressures of those 17 congeners vary from 5 × 10-10 to 2 × 10-7 Pa at 25 °C (1) and decrease as chlorination level increases. A relevant study (2) indicates that ambient PCDD/Fs in Taiwan originate mainly from waste incineration processes including municipal waste incinerators (MWIs), medical waste incinerators, and industrial waste incinerators (IWIs). Over 16 000 tons of municipal wastes and 7000 tons of industrial wastes are incinerated each day in Taiwan. In addition to the waste incineration processes, major anthropogenic sources for PCDD/F emissions include industrial processes such as chemical manufacturing and metal smelting processes. In flue gases, PCDD/Fs may exist in gaseous * Corresponding author tel and fax: 886-3-4226774; e-mail: [email protected]. 10.1021/es0501722 CCC: $30.25 Published on Web 09/07/2005

 2005 American Chemical Society

form (vapor phase) or be bound to particulate matter (solid phase). To meet the stringent PCDD/F emission standards, PCDD/F emission sources are generally equipped with various types of air pollutant control devices (APCDs), leading to different levels of PCDD/F control. Our previous study (3) demonstrated that partitioning of PCDD/Fs between the vapor and solid phases changes significantly as the flue gas passes through different APCDs. The key parameters controlling the phase variation include the congener’s vapor pressure, particle concentration, and removal mechanism of the APCDs applied. Vapor-phase PCDD/Fs can be emitted from the stack by penetrating through the APCDs which are designed only for controlling particulate matter such as cyclone (CY), bag filter (BF), and electrostatic precipitator (EP). Vapor-phase PCDD/Fs can be removed by various means including adsorption with carbon-based adsorbents and catalytic destruction. Among the methods available for use, spraying of powdered activated carbon (PAC) into gas streams or installing a fixed-bed system to adsorb PCDD/Fs are considered the simplest. Nevertheless, some problems exist in controlling PCDD/F emissions from waste incinerators equipped with activated carbon injection (ACI). The removal efficiencies of PCDD/F congeners achieved with ACI are not always consistent due to the variation of vapor pressure and different adsorbing capacities of the activated carbons for different congeners (4). Normally, lowly chlorinated PCDD/Fs are mostly distributed in vapor phase and can be readily adsorbed by activated carbon. Because of the low vapor pressure of highly chlorinated PCDD/Fs, over 70% of those congeners in the flue gas are distributed in solid phase. In general, activated carbon adsorption systems can only effectively remove vapor-phase pollutants. Compared to lowly chlorinated PCDD/Fs, highly chlorinated PCDD/Fs with lower vapor-phase partitioning are harder to adsorb with activated carbon. Hence, the removal efficiency achieved with activated carbon adsorption for each PCDD/F congener decreases with the increase of the solid-phase partitioning. To collect sufficient mass of particulate for accurate analysis of the concentration, it is often necessary to sample in a stack for 3 h or longer. No real-time analyzer is currently available for instantaneous measurement of PCDD/F concentration of flue gases. For these reasons, vapor and solid partition based on stack sampling data is highly uncertain. Limited studies have been completed so far to compare the phase distributions of PCDD/F congeners based on flue gas samplings. Cavallaro et al. (6) was one of the first investigators to interpret the vapor/solid partitioning with a sampling train. They observed that the PCDD/F emissions from MWIs seemed to predominate in vapor phase, and attributed it to the relatively high temperatures of stack gases during sampling which might have promoted desorption of PCDD/ Fs from particles. Another research study (7) indicated that approximately 65% of the PCDD/Fs were found in vapor phase at the furnace exit and increase to 85% at the inlet of APCDs. Although the flue gas temperature at the furnace exit was higher than that at the inlet of APCDs, the vaporphase PCDD/Fs at the furnace exit was actually lower than that measured at the inlet of APCDs. The study excludes the possibility that most vapor-phase PCDD/Fs was generated by de novo synthesis. Hence, information on which form of PCDD/Fs exist in flue gases is important in selecting and designing PCDD/F control equipment. To examine this important feature, this study was motivated to investigate the partitioning of PCDD/Fs between vapor/solid phases of stack gas. We focus on the understanding of the partitioning and removal efficiency of PCDD/Fs of flue gas at one VOL. 39, NO. 20, 2005 / ENVIRONMENTAL SCIENCE & TECHNOLOGY

9

8023

TABLE 1. Condition of Flue Gas at Different Sampling Points in Four Incineratorsa MWI DSI+ACI+BF inlet capacity (tons/day/incinerator) flue gas flow rate (kNm3/hr/incinerator) temperature (°C) CO2 (%) O2 (%) particulate matter (PM) concentration (mg/Nm3) PM removal efficiency (%) a

IWI-1 stack

SDA+ACI+BF inlet

450 96.6 202 10.9 9.9 770

IWI-2 FCB inlet

stack

96 27 138 8.6 11.2 1.2

99.8

346 8.1 12.0 9521 99.9

stack 60 10

161 5.2 14.8 5.1

60 8.6 13.1 145.8

75 8.0 13.6 65.8 54.9

DSI: dry sorbent injection. SDA: spray dryer absorber. ACI: activated carbon injection. BF: bag filter. FCB: fixed activated carbon bed.

FIGURE 1. Flow diagram and sampling points of three incinerators investigated. municipal wastes incinerator (MWI) and two industrial waste incinerators (IWI-1 and IWI-2) equipped with different types of activated carbon systems as air pollution control devices (APCDs).

Materials and Methods Sampling Sites. The municipal waste incinerator (MWI) investigated in this study started to operate in 1995. It consists of three incinerating units, each with its own heat recovery system. This MWI is equipped with cyclone (CY), dry lime sorbent injection systems (DSI), and bag filters (BF) for controlling acid gas and particulate emissions. As high as 4.5 ng-TEQ/Nm3 of PCDD/F concentrations were measured in the stack gas of this MWI back in 1998 (8). In 1999 Taiwan promulgated PCDD/F emission limits for existing large-scale MWIs (0.1 ng-TEQ/Nm3) to reduce PCDD/F emissions. The ACI technology was retrofitted in the investigated MWI in March 1999 for reducing PCDD/F emissions to meet the stringent standards. The IWI-1 investigated in this study started to operate in 1995. It is a fluidized bed incinerator and has the capacity of 4 tons per hour. IWI-1 was equipped with a spray dryer absorber (SDA) and bag filters (BF) for controlling acid gas and particulate emissions. The IWI-1 has been retrofitted with the ACI system for reducing PCDD/F emissions to meet the standard since 2001. On the other hand, IWI-2 is equipped with a venturi scrubber, wet electrostatic precipitator (WEP), and fixed carbon bed (FCB) as major APCDs. The capacity of the IWI-2 investigated in this study is 2.5 tons per hour. The operating and flue gas conditions of the MWI and two industrial waste incinerators (IWI-1 and IWI-2) investigated in this study are shown in Table 1. In addition, the characteristics of the ACI system and FCB system used in those three incinerators are listed in Table 2. The gas flow sheets and PCDD/F sampling points of the three incinerators investigated in this study are schematically shown in Figure 1. Sample Collection. In this study, 24 vapor/solid-phase PCDD/F samples were collected at sampling points of 8024

9

ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 39, NO. 20, 2005

TABLE 2. Characteristics of the ACI and FCB System Used in Three Incinerators Investigated material form injection rate surface area (m2/g) diameter distribution (mm)

MWI

IWI-1

IWI-2

coconut shell

coconut shell

powdered 50 mg/Nm3 >1050 1000

bituminous coal granular fixed bed >1200 3-4

activated carbon system (ACI and FCB) inlet and stack in three incinerators investigated. The flue gas sampling was conducted with the Graseby Anderson stack sampling system complying with U.S. EPA Method 23. The vapor-phase sample was collected by XAD-2 resin while the particle-bound portion was collected by glass fiber filter and by rinsing of the sampling probe thereafter. The glass fiber filter (WhatmanEPM 2000 110 mm) used in this study collects test particles with mean diameter of 0.3 µm with collection efficiency exceeding 99.95% at the flow rate 1.5 mg/cm2) during the flue gas sampling would significantly decrease the partitioning of vapor-phase PCDD/Fs by 60%. The filter loading was controlled to be lower than 1.4 mg/cm2 by replacing the filter in this study. Hence, the partitioning of solid-phase PCDD/Fs and highly chlorinated PCDD/Fs would not be overestimated and the

FIGURE 2. Variation of PCDD/F concentration in vapor/solid phases at different sampling points in MWI, IWI-1, and IWI-2. bias caused by the high particle loading at the filter is minimized. Sample Analysis. Once the flue gas sampling was completed, the samples were brought back to the laboratory under refrigeration. They were then spiked with known amounts of U.S. EPA Method 23 internal standard solution. Thereafter, the XAD-2 and filter sample were Soxhlet extracted with toluene for 24 h. The toluene extract was then concentrated to about 1 mL by rotary evaporation and was replaced by 5 mL of hexane for the pretreatment process. Having been treated with concentrated sulfuric acid, the sample was then subjected to a series of cleanup columns including sulfuric acid silica gel column, acidic aluminum oxide column, and Celite/carbon column. Finally, the recovery standard solutions were spiked with known amounts of Method 23 internal standard solutions, and then analyzed for 17 2,3,7,8substituted PCDD/F congeners with high-resolution gas chromatography (HRGC; Hewlett-Packard 6890 plus)/highresolution mass spectrometer (HRMS; JEOL JMS-700D) equipped with a fused silica capillary column DB-5 MS (60 m × 0.25 mm × 0.25 µm, J&W).

A field blank, a laboratory blank, and a matrix spike sample (2.0-20 pg/µL) were incorporated in the analytical procedure for every eight samples for the purpose of quality control. Detection limits (0.02-0.2 pg) were derived from the blanks and quantified as 3× the standard deviations of the mean concentration in the blanks. In this study, the concentrations of all field blank and laboratory blank samples were lower than 30 pg. Furthermore, the mean recoveries of standards for all 13C12-2,3,7,8-substituted PCDD/Fs range from 61 to 119%, which are within the acceptable 40-130% range set by the U.S. EPA Method 23.

Results and Discussion Average PCDD/F Concentrations and Congeners Distribution in Flue Gas of Four Waste Incinerators. Figure 2 shows the average PCDD/F concentrations in flue gases at different sampling points. Results of the flue gas sampling indicate that the average PCDD/F concentrations are 3.93, 19.6, and 6.05 ng-TEQ/Nm3 at DSI+ACI+BF, SDA+ACI+BF, and FCB inlets, respectively. Besides, the average PCDD/F concentrations in stack gases are 0.17, 0.014, and 1.74 ng-TEQ/Nm3 for VOL. 39, NO. 20, 2005 / ENVIRONMENTAL SCIENCE & TECHNOLOGY

9

8025

MWI, IWI-1, and IWI-2, respectively. On the basis of the PCDD/F analysis method (NIEA.807.70C) adopted in Taiwan, the PCDD/F concentration in flue gas is expressed as ngTEQ/Nm3, in which the normal (N) state refers to 1 atm, 0 °C (273.15 K), and corrected for 11% oxygen content. In MWI, the PCDD/F concentration measured in stack gas was slightly higher than the PCDD/F emission limit (0.1 ng-TEQ/Nm3) adopted for large-scale MWIs in Taiwan (Figure 2). Nevertheless, if we compare the results obtained in this study with the results obtained in 1999, the ACI technology effectively reduces PCDD/F concentration emission. In IWI-1, the PCDD/F concentration in flue gas at the SDA+BF inlet was significantly higher than that at other sampling points. The relatively high PCDD/F concentration measured at this point was attributed to the fact that there are no APCDs installed prior to the SDA+BF inlet in IWI-1. Hence, the particle and solid-phase PCDD/F concentration in IWI-1 is the highest of the three incinerators investigated. Evereart and Baeyens (11) indicate that de novo synthesis taking place between 200 and 400 °C results in a much higher PCDF/PCDD ratio (generally greater than 1). In IWI-1, the temperature of flue gas at SDA+BF inlet was 346 °C which was within the de novo synthesis temperature window and the PCDF/PCDD ratio in the flue gas was 3.1. Hence, the high PCDD/F concentration measured at SDA+BF inlet in this study was more likely attributed to the de novo synthesis (12). In IWI-2, the PCDD/F concentration in stack gas was higher than the PCDD/F emission limit (0.5 ng-TEQ/Nm3) adopted for smallscale IWIs in Taiwan. Undoubtedly, a more effective technology is needed to further reduce its PCDD/F emissions to comply with the regulation. It is noted that the solid-phase PCDD/F concentration increases slightly as the flue gas passes through the FCB. This issue will be discussed in the next section. Figure 3 shows PCDD/F congener distributions in flue gases collected at different sampling points in MWI, IWI-1, and IWI-2, respectively. In MWI, PCDFs accounts for about 67% of total PCDD/Fs at the DSI+BF inlet, while PCDFs accounts for about 76% and 79% of PCDD/Fs measured at the SDA+BF and FCB inlet in IWI-1 and IWI-2, respectively. The difference may be attributed to the variation of the input wastes (municipal versus industrial wastes) between MWI and IWIs (IWI-1 and IWI-2). Besides, Figure 2 also indicates that a relatively high PCDD/F concentration was measured at the incinerator outlet in IWI-1. This is attributed to the fact that the flue gas temperature at that sampling point is 346 °C which is within the de novo synthesis temperature window, resulting in a much higher PCDF/PCDD ratio (generally greater than 1). PCDFs account for over 75% of total PCDD/Fs. As the flue gas passes through the ACI+BF, the vapor- and solid-phase PCDD/Fs are effectively removed. Hence, the major shift in PCDD/F congener distribution for IWI-1 is attributed to the formation of the de novo synthesis and the removal mechanism achieved with ACI+BF. However, a unique feature was observed for PCDD/F distribution measured across the FCB in IWI-2. The distribution of lowly chlorinated PCDF congeners (especially PeCDF) in flue gas downstream of FCB is significantly higher than that of all PCDD congeners measured (even higher than OCDD). Comparison of PCDD/F Removal Efficiency with Different APCDs. Figure 4 shows the PCDD/F removal efficiencies in flue gases achieved with two different activated carbon systems (ACI and FCB). The average removal efficiencies are 95%, 99.8%, and 65.3% for MWI, IWI-1, and IWI-2, respectively. Besides, the removal efficiencies of solid-phase PCDD/F congeners do not change significantly in the MWI and IWI-1 investigated. On the other hand, removal efficiencies of vapor-phase PCDD/Fs decrease with increasing chlorination (especially of PCDD) for the incinerators equipped with ACI+BF (for MWI and IWI-1). The removal 8026

9

ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 39, NO. 20, 2005

mechanism of solid-phase PCDD/Fs relies on filtration of the bag filter (average removal efficiency for particulate matter is over 99.8%). As expected, the solid-phase PCDD/F and particle removal efficiencies achieved with bag filter are fairly close (98.5%-99.8% and 99.8%-99.9%, respectively). In addition, activated carbon could adsorb volatile organic pollutant effectively, lowly chlorinated congeners are of higher vapor pressures compared to highly chlorinated congeners and have higher tendencies to exist in vapor phase and be adsorbed by activated carbon. A previous study (13) indicates that vapor-phase PCDD/F removal efficiency achieved with ACI increases with increasing injected PAC concentration. Hence, the higher PAC injection rate in IWI-1 (120 mg/Nm3) results in a higher vapor-phase PCDD/F removal efficiency compared with that achieved in MWI in which a lower PAC injection rate (50 mg/Nm3) is applied. Surprisingly, the results indicate that solid-phase PCDD/F particle removal efficiency is negative for IWI-2. The solidphase PCDD/F concentration increases slightly as the flue gas passes through FCB. This is possibly attributed to the attrition of carbon particles within the FCB. Particulate matter collected at different sampling points in IWI-2 has been analyzed for carbon content. Figure 5 shows that the carbon content in particles and the carbon mass concentration in flue gas increase (about 1.5-3 times) as the flue gas passed through the FCB, confirming the hypothesis of carbon attrition which results in the increase of the solid-phase PCDD/F concentration as flue gas passes through FCB. ACI technology can effectively remove vapor-phase PCDD/Fs while BF is effective in removing solid-phase PCDD/Fs in both MWI and IWI-1. The removal mechanism of solid-phase PCDD/Fs relies on filtration of bag filter. The solid-phase PCDD/F and particle removal efficiencies achieved with bag filter are fairly close. Besides, partitioning of PCDD/Fs between vapor and solid phases in flue gas after PCDD/Fs control devices (ACI and FCB) is quite different. In general, FCB could not remove solid-phase PCDD/Fs, and the small particles in flue gas are harder to collect with APCDs (WEP and FCB) and may be richer in carbon content. We speculate that the increasing solid-phase PCDD/F concentration measured at FCB outlet in IWI-2 is possibly attributed to the attrition of granular activated carbon within the FCB and penetration of small particles with higher carbon content through WEP and FCB. Hence, a slight increase of solid-phase PCDD/Fs is observed in the stack gas. Vapor/Solid-Phase Partition of PCDD/F Congener in Flue Gas of Three Incinerators. Figure 4 indicates that the vapor/solid-phase PCDD/F removal efficiencies vary with the chlorination level of each PCDD/F congener. To evaluate the partitioning of 17 2,3,7,8-substituted PCDD/F congeners between vapor and solid phases, the vapor-solid distribution coefficient for each congener was calculated from eq 1. φ is used to represent the coefficient of vapor/solid-phase PCDD/F congeners partition in logarithm. When the φ value is >0, it indicates that over 50% of the PCDD/F congeners are distributed in vapor phase.

φ ) log

() Cv Cs

(1)

where φ is coefficient of vapor/solid-phase PCDD/Fs partition; Cv is concentration of PCDD/Fs congeners existing in vapor phase (ng/Nm3); and Cs is concentration of PCDD/Fs congeners adsorbed on particles (ng/Nm3). Previous study indicates (14) that vaporization is the major mechanism that causes solid-phase PCDD/Fs to transfer into the vapor phase (especially for the congeners of lowly

FIGURE 3. Characteristics of PCDD/F congener distribution at different sampling points in MWI, IWI-1, and IWI-2. chlorinated level, like TCDD). Hence, the coefficient of semivolatile compounds (like PCDD/Fs) adsorbed to particles was mainly affected by the vapor pressures of those compounds. On the other hand, the vapor pressures of PCDD/F congeners increase as the gas temperature increases. Eitzer and Hites (15, 16) have correlated saturation vapor pressure (PLo) of PCDD/Fs with gas chromatographic retention indexes (GC-RI) on a nonpolar (DB-5) GC column using p,p′-DDT as a reference standard. The correlation has been modified using

RI (retention index) developed by Donnelly et al. (17) and Hale et al. (18).

-1.34(RI) 1320 +1.67 × 10-3(RI) +8.087 T T

logPLο )

(2)

where PLο is saturation vapor pressure of the organic compound (Pa); RI is retention index (from 2338 to 3196 for each of 17 2,3,7,8-substituted PCDD/F congeners); and T is temperature (K). VOL. 39, NO. 20, 2005 / ENVIRONMENTAL SCIENCE & TECHNOLOGY

9

8027

FIGURE 4. Removal efficiencies of PCDD and PCDF congeners in vapor and solid phases achieved with DSI+ACI+BF, SDA+ACI+BF, and FCB, respectively, in three incinerators.

FIGURE 5. 5. Carbon content of particulate matter and carbon concentration in flue gases at different sampling points in the IWI-2 investigated. To compare the partitioning of PCDD/Fs between vapor and solid phases in flue gases with different APCDs, data collected are further analyzed. Figure 6a and b show the trend between the log(Cv/Cs) versus log PLο of each PCDD/F congener in flue gases downstream of the traditional APCDs (such as CY or EP) and PCDD/F control devices (such as 8028

9

ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 39, NO. 20, 2005

ACI+BF or FCB), respectively. Each dot represents the log(Cv/Cs) versus log PLο of each of 17 2,3,7,8-substituted PCDD/F congeners in the flue gas operating at different temperatures. To investigate the partitioning of PCDD/Fs between vapor and solid phases with different APCDs, we refer to the results (3) obtained from another large-scale municipal wastes incinerator (MWI*) which is equipped with an electrostatic precipitator (EP), a wet scrubber (WS), and a selective catalytic reduction system (SCR) as major APCDs. In Figure 6a, the values of log(Cv/Cs) are all >0 in flue gases downstream the particle control devices in MWI, IWI-2, and MWI*. This is attributed to the fact that the solid-phase PCDD/Fs are effectively removed by CY, WEP, and EP, hence PCDD/F congeners are mostly distributed in the vapor phase (over 70%). In IWI-1, no APCD is available prior to the sampling point, and the log(Cv/Cs) values are mostly