Environ. Sci. Technol. 2003, 37, 1219-1224
Entrained-Phase Adsorption of PCDD/F from Incinerator Flue Gases K. EVERAERT,* J. BAEYENS, AND J . D E G R EÅ V E Katholieke Universiteit Leuven, Department of Chemical Engineering, de Croylaan 46, B-3001 Heverlee, Belgium
The emission abatement of polychlorinated dioxins and furans (PCDD/F) issued from municipal solid waste incineration (MSWI) is growing in importance because of more stringent emission standards and general health concern. These substances cannot be separated by conventional gas cleanup processes. They are successfully removed through adsorption onto carbonaceous materials, and the entrained-phase injection of pulverized adsorbents in the flue gas, followed by high-efficiency separation, is widely applied. Operating conditions and results obtained in Flemish MSWIs are given. The results illustrate the excellent overall removal efficiency: the regulation limit of 0.1 ng TEQ/Nm3 dry gas at 11% O2 can be achieved. Furans are adsorbed to a slightly higher extent than the dioxins. The PCDD/F removal by carbonaceous adsorbents is thereafter modeled from first principles for the contribution of both entrained-phase (η1) and cake filtration (η2) to the overall efficiency (ηT), with dominant parameters being the operating temperature, the dosage and activity of adsorbent, and the fraction of adsorbent in the filter cake. Application of the model equations and comparison of measured and predicted overall efficiencies for the Flemish MSWIs demonstrate the validity of the model, which enables the MSWI operators both to predict the adsorption efficiencies for combinations of major operating parameters and to assess the sensitivity of the process to varying operating conditions. Finally, some practical difficulties encountered with the entrained-phase adsorption are discussed.
Introduction The emission abatement of toxic chemicals, such as polychlorinated dioxins and furans (PCDD/F) as well as polychlorinated benzenes and phenols, polycyclic aromatic hydrocarbons (PAHs), and some heavy metals, is growing in importance because of general environmental and health concern and is reflected in more stringent emission standards for these components. Prior to meeting the 0.1 ng TEQ/Nm3 regulations, the incineration of municipal or hospital waste was a major PCDD/F emission source. Current incinerator emissions of these components (below 0.1 ng TEQ/Nm3) are small in comparison with the overall emission rates of PCDD/F to the environment (1-4). These toxic pollutants are now also being targeted in metallurgical processes (e.g., sintering, founderies) (e.g., refs 2 and 5), chemical plants (e.g., cracking units, reformers), coal-fired power stations (e.g., refs 5 and 6), and open-burning/house firing. * Corresponding author telephone: 00 32 16 32 2687; fax: 00 32 16 32 2991; e-mail:
[email protected]. 10.1021/es020020w CCC: $25.00 Published on Web 02/14/2003
2003 American Chemical Society
These groups of substances cannot be separated by conventional gas cleanup processes. A suitable process is the adsorption by carbon-containing adsorbents such as activated carbon, lignite, or coke (5, 7-13) applied either through the injection of an adsorbent and subsequent removal by filtration or in a separate moving/fixed-bed adsorber at the end of the flue gas cleaning trail. The simplest technique is the injection of pulverized sorbents in the flue gas, thus avoiding high investment costs of separate adsorption systems. This paper specifically deals with the abatement of PCDD/ F. Numerous studies (summarized in refs 2 and 6) have demonstrated that PCDD/F are formed in cooler sections of the incineration plant through pyrocondensation from precursors or through de novo synthesis with the aid of the catalytic action of flue gas particles. At the common temperature range after the boiler, approximately 50-90% of the PCDD/F will be present in the fly ash (6, 10). This fraction can be removed through adequate dust filtration (mostly using an electrostatic precipitator, ESP), although dust and other pollutant emission standards cannot generally be met using this step alone. The remaining PCDD/F are believed to be mostly in the gaseous state. To meet the severe emission standards, adsorbent is injected into the flue gas stream and collected in a high-efficiency filtration step, either a multiple-field ESP or a fabric filter unit, which is mostly applied when retrofitting existing MSWIs. The present paper describes PCDD/F abatement results obtained in Flemish MSWIs that because of the retrofitting concept mostly use an entrained-phase adsorption and subsequent fabric filter separation. The results will illustrate the overall removal efficiency as well as the adsorption behavior of the different PCDD/F congeners. It will show that the injection of adsorbents into the flue gas and subsequent removal through a fabric filter can achieve the required PCDD/F removal rates. Design equations will be presented.
Evaluation of the MSWIs in Flanders Entrained-Phase Adsorption Concept. The adsorption of PCDD/F [and other pollutants, e.g., mercury, PAHs, polychlorinated aromatics etc. (7-11, 14, 15)] requires that the sorbent be brought into contact with the waste gas to be cleaned. Various processes are available at present (8), ranging from the moving (packed) bed filters to the adsorption in an entrained dust cloud with a downstream electrostatic precipitator or fabric filter to remove the spent sorbent and possibly residual process dust from the waste gas. These solutions are illustrated in ref 8. The post-ESP (add-on) technique of injection and subsequent additional fabric filter was used when retrofitting existing MSWIs in Flanders. In the entrained-phase adsorption, the pulverized sorbent is homogeneously injected into the waste gas and flows cocurrently to a filtration step. A sufficiently long residence time in the gas ducting from the injection point to and in the filter should be guaranteed. This technique is mainly used as a nonselective safety filter for trace separation. The fineness and characteristics of the sorbent are of paramount influence on the adsorption efficiency. Applied Techniques in the Flemish MSWIs. In 1993, Flanders operated 19 MSWIs, none of them equipped with specific dioxin-abatement systems. Five incinerators were shut down as a result of the initial monitoring campaign (16) because of either the use of obsolete and/or inadequate technology or the resulting excessive pollutant emissions. The other MSWIs added necessary best-available technology VOL. 37, NO. 6, 2003 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
1219
TABLE 1. MSWIs in Flanders: Waste Gas Treatment Systemsa capacity (no. of furnaces × unit plant capacity in t/h)
4 5 6 7 8
2 × 6.2 2 × 10 2 × 3.6 2 × 1.8 2 × 5.6 2 × 2.0 2 × 5.5 2 × 4.0 2 × 5.5
9 10 11
3 × 9.0 2 × 4.0 2 × 7.5
12
3 × 20
1 2 3
status 1994
status 2002
ER, ESP CT, ESP ER, ESP, 2sS CT, ESP ER, ESP, 2sS ER, CYC CT, ESP ER, ESP ER, ESP
ER, ESP, REAC, FF, 2sS ER, ESP, REAC, FF, 2sS ER, ESP, REAC, FF, 2sS closed in 1998 ER, ESP, REAC, FF, 2sS ER, REAC, cFF ER, ESP, REAC, FF, 2sS ER, REAC, FF ER, ESP, REAC, cFF, 2sS, rH, SCR ER, VS ER, VS, aS, rH, Inj, FF ER, ESP, REAC, FF ER, ESP, REAC, cFF CT, ESP CT, ESP, ER, REAC, FF, 2sS commissioned in ER, REAC, FF 1998
a
CT, cooling tower; CYC, multi-cyclones; cFF, catalytic fabric filter (without adsorbent); ER, energy recovery; ESP, electrostatic precipitator; FF, fabric filter; Inj, injection of adsorbent, without contacting vessel; rH, reheating of flue gas; SCR, catalytic deNOx; aS, alkaline scrubbing stage; VS, venturi scrubber (high-pressure and acid-scrubbing stage); 2sS, two-stage scrubber; REAC, injection of (mostly) Ca(OH)2 and adsorbent (activated carbon, lignite) and reaction in an in-line contacting vessel.
(BAT) to meet the stringent emission standards that came into force in 1997. A further three (smaller) units operated till 1998. A new plant of 200 000 t/yr was commissioned in 1998 and expanded to 350 000 t/yr in 2001. All currently operating plants are summarized in Table 1 as far as the waste gas treatment systems are concerned. These systems are combinations of several unit processes and involve mostly the injection of alkaline sorbents [Ca(OH)2, Sorbalit] and activated carbon or lignite, followed by baghouse filtration and two-stage scrubbing (acid and alkaline). Two plants utilize catalytic fabric filters to destroy PCDD/F compounds and, hence, require no injection of adsorbent. In view of the future reduction of the NOx standards, several plants already utilize the selective noncatalytic reduction principle (SNCR) for NOx abatement. One plant uses the selective catalytic reduction principle (SCR). The stack emission levels are measured both by a continuous flue gas sampling/adsorption (2 weeks) and subsequent analysis and by the periodic traditional 8-h sampling method and subsequent analysis (17). The PCDD/F stack emission levels are very low: below 0.1 ng TEQ/Nm3 dry at 11% O2 for the continuous sampling and always significantly below the 0.1 ng limit for the 8-h probing. Some important operational factors are as follow: (i) The injection rate of Ca(OH)2 expressed as “g of CaO/ Nm3” ranges from 2 to 4 g/Nm3 flue gas and is function of the SO2 and HCl/HF concentrations in the flue gas. (ii) The injection rate of adsorbent ranges from 50 to 80 mg of C/Nm3 for activated carbon or 200-400 mg/Nm3 for pulverized coke. (iii) The applied face velocity of the fabric filter ranges from 0.008 to 0.017 m/s with an average value of 0.011 m/s. (iv) The applied flue gas velocity in the ducting between the reactor and the fabric filter varies from 0.8 to 4 m/s with an average of 1.8 m/s. (v) The contact time of the flue gas and the adsorbent prior to the baghouse filter ranges from 0.8 to 7 s, with an average of 2.6 s. (vi) The fraction of the adsorbent (carbon) in the filter cake varies in function of the applied waste gas treatment cycle and lies between 0.5 and 2% as a function of the dust abatement yield of the primary dedusting. 1220
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 37, NO. 6, 2003
TABLE 2. PCDD/F Removal Efficiency in Flemish MSWIs ref
inlet PCDD/F concn (ng TEQ/Nm3 dry at 11 vol % O2)
PCDD/F removal efficiency, ηT (%)
1 2 3 4 5a 6 7 8a 9 10a 11 12
2.7-2.9 5.1-7.3 2.6-4.2 6.1-8.4 6.0-15.8 5.3-10.9 2.3-9.2 1.0-3.8 0.2-1.4 0.3-1.3 3.3-12.3 1.6-4.1
96.3-98.3 98.2-99.3 96.2-98.8 98.4-99.4 98.3-99.4 98.1-99.1 97.8-99.0 95.4-98.3 93.6-96.4 93.3-98.8 98.4-99.3 96.9-98.7
a
Nonadsorption abatement.
(vii) The temperature of the flue gas at the level of the fabric filter is generally 140-160 °C, with higher values (220240 °C) when catalytic fabric is used. (viii) In most of the plants, the pulsed back-cleaning of the fabric filter is controlled by the accepted pressure drop of the cake. Cycle durations of approximately 2 h (e.g., unit 12) to a maximum of 10 h (e.g., unit 2) have been indicated.
Results and Discussion of MSWI Data Generalities. The full picture of the 12 Flemish MSWIs is available both before and after PCDD/F abatement. The concentrations after the abatement do not differ from stack emissions: the effect of the possibly used final two-stage scrubber on PCDD/F emission levels is insignificant. Plants 5, 8, and 10 are not discussed in the paper since they used a Goretex catalytic filter (plants 5 and 10) or a Goretex SCR filer (plant 8) as the final abatement process. At present, PCDD/F stack emissions are far below the standard of 0.1 ng TEQ/Nm3 dry at 11% O2, with the concentration of various congeners even below their respective detection limits. An analysis was made of over 300 available data sets for the periods 1993-1994 and 19982001. The following average PCDD/F inventory can be made: Overall PCDD/F Removal Efficiency. As stated before, some congeners have concentrations below the detection limit, thus making the calculation of the overall removal efficiency rather imprecise. Removal percentages are given on ng TEQ/Nm3 dry at 11 vol % O2 basis. The efficiency is calculated as the ratio of the TEQ reduction and the inlet TEQ value. Table 2 depicts reported maximum and minimum inlet concentrations and associated removal efficiencies. The data illustrate that the removal efficiency is always greatly in excess of 93%. The concentrations of Table 2 moreover illustrate that the outlet concentration is on average 50 times lower than the inlet concentration (Cout/Cin ) 1 - ηT). “Fingerprint” Analysis for MSWI Flue Gases. Table 3 shows the average PCDD/F fingerprint, typical for the Flemish adsorbent-treated MSWI flue gases both prior to and after abatement. These average data are also illustrated in Figure 1 for several plants having experimented at the authors’ request with the added quantity of activated carbon. From Table 3 and Figure 1, it is clear that (i) The degree of chlorination remains fairly constant, with hepta and octa congeners being dominant. (ii) The addition of activated carbon does not significantly modify the congener fingerprint, although furans tend to be removed more efficiently than dioxins as shown by the decreasing furan/dioxin ratio. Similar observations were reported by Tejima et al. (15). (iii) The increased quantity of 50-200 mg of AC/Nm3 (AC, activated carbon) has a limited effect on the overall PCDD/F
TABLE 3. Review of Fingerprint Data for Flemish MSWIs chlorination degree
wt % of congeners
prior to abatement [SD] after abatementa [SD] a
TCDD
PeCDD
HxCDD
HpCDD
OCDD
TCDF
PeCDF
HxCDF
HpCDF
OCDF
F/D ratio
PCDD
PCDF
0.2 [0.03] 0.08 [0.002]
0.9 [0.4] 0.5 [0.07]
4.5 [1.7] 2.3 [0.1]
12.9 [4.2] 12.7 [0.4]
20.3 [9.9] 37.6 [10]
1.5 [0.7] 1.2 [0.7]
6.5 [3.6] 3.3 [1.2]
15.9 [6.5] 9.9 [3.4]
24.7 [4.2] 19.2 [2.2]
12.7 [4.8] 13.3 [1.8]
1.7 [0.7] 0.91 [0.3]
7.35 [0.1] 7.64 [0.07]
6.69 [0.1] 6.87 [0.13]
Values are approximate since various congeners are below their respective detection limit.
FIGURE 1. Effect of activated carbon on PCDD/F fingerprint.
TABLE 4. Powder Sorbent Characteristics Rotary Hearth cokes (Rheinbraun) median size (µm) bulk density (kg/m3) iodine no. micropores 99%, it is impossible to effectively compare them with the model predictions because of the lack of values for the necessary underlying operating conditions. The overall efficiency results of Ruegg and Sigg (13) however were used to quantify some model parameters. The paper presents ηT for a given dosage of activated carbon at temperatures from 117 up to 121 °C. Using estimated operating conditions for parameters tt, vf, vt, and d (in line with average values for Flemish MSWIs), the model equations fit the measured results for fA ≈ 1% and fs ≈ 15-20%. These values correspond with the range of previously accepted values for the model assessment, again stressing the predictive quality of the model equations.
Practical Considerations and Recommendations Although the entrained-phase adsorption is a relatively simple technique, easily implemented in new and retrofit MSWIs, there are some drawbacks and difficulties encountered: (i) The de novo synthesis from a carbon source has been previously demonstrated and is enhanced at higher temperatures, with a maximum formation rate at about 300 °C (1-5). In analogy with the PCDD/F concentrations encountered in fly ash (6), it is expected that the formation at 150 °C is 1.0% only of the levels at 270 °C. It is hence obviously important that AC should be injected in cooler flue gases, and temperatures of 130-160 °C are recommended and commonly used. (ii) Adsorption moves the gaseous PCDD/F into the solid phase and does not reduce the overall environmental burden. This is achieved in the catalytic destruction of PCDD/F, which should therefore be considered a better abatement technique
1224
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 37, NO. 6, 2003
for gaseous PCDD/F once proven reliable and efficient during long periods of operation. Since 60-90% of the PCDD/F are always adsorbed on the fly ash, the overall reduction effect of the catalytic destruction remains moderate. (iii) The use of fine carbon powder implies safety hazards, as illustrated by explosion characteristics of Table 2. Specific measures, such as temperature measurements complemented by, for example, ∆CO monitoring in storage silos, are required. (iv) Although entrained-phase adsorption is easy to use in new or retrofit MSWIs, the operating costs are fairly high: if injected at 50 mg/Nm3 in an average flue gas flow rate of 4000 Nm3 h-1 (t of solid waste)-1, the required dosage is 40 t/yr for a 200 000 t/yr MSWI. The use of coke or lignite might offer economic advantages despite the higher dosage required.
Literature Cited (1) Everaert, K.; Baeyens, J. J. Air Waste Manage. Assoc. 2001, 51 (5), 718-724. (2) Everaert, K.; Baeyens, J. Proceedings of 6th World Congress of Chemical Engineering, Melbourne, September 23-27, 2001. (3) Donghoon, S.; Sangmin, C.; Jeong-Eun, O.; Yoon-Seok, C. Environ. Sci. Technol. 1999, 33 (15), 2657-2666. (4) Sierhuis, W. M.; de Vries, C.; Born, J. P. G. Chemosphere 1996, 32 (1), 159-168. (5) Kinzel, J.; Gebert, W.; Gara, S. Organohalogen Compd. 1994, 19, 311-316. (6) Everaert, K.; Baeyens, J. Chemosphere 2002, 46 (3), 439-448. (7) Yamaguchi, H.; Ogaki, Y.; Nakamura, S.; Okuyama, K.; Shibuya, E.; Nomura, T. Organohalogen Compd. 1994, 19, 411-414. (8) Esser-Schmittmann, W.; Lenz, U.; Erken, M. Publication Rheinbraun AG, Ko¨ln(D), 1996. (9) Lu ¨ der, K. Presented at the Dioxin Conference, Vrije Universiteit Brussel, Brussels, September 1996. (10) Wirling, J.; Esser-Schmittmann, W.; Lenz, U. Presented at the Dioxin Conference, Vrije Universiteit Brussel, Brussels, September 1997. (11) Egeler, R.; Seitz, B. Presented at the Dioxin Conference, Vrije Universiteit Brussel, Brussels, September 1998. (12) Hahn, J. Chemosphere 1992, 25 (1-2), 57-60. (13) Ruegg, H.; Sigg A. Chemosphere 1992, 25 (1-2), 143-148. (14) Blumbach, Y.; Nethe, J. P. Organohalogen Compd. 1994, 19, 305-310. (15) Tejima, H.; Nakawage, I.; Shinoda, T.; Maeda, I. Chemosphere 1996, 32 (1), 169-175. (16) Baeyens, J.; Serruys, H.; Bernaert, P. Report available from the Flemish Ministry of the Environment, Division of Environmental Inspection, de Ferraris Building, Brussels (B), D/1994/6137/42, 1994. (17) Hagenmaier, H.; Ho¨ckel, J. Organohalogen Compd. 1995, 22, 437-438. (18) EAB Control Cost Manual, 3rd ed.; U.S. Environmental Protection Agency: Research Triangle Park, NC, 1987. (19) Perry, R. H.; Green, D. Perry’s Chemical Engineers’ Handbook, 6th ed.; McGraw-Hill: New York, 1984; Chapter 16. (20) Milligan, M. S.; Altwicker, E. R. Environ. Sci. Technol. 1996, 30 (1), 225-236. (21) Baeyens, J.; Van Gauwbergen, D.; Vinckier, I. Powder Technol. 1995, 83, 139-148. (22) Kunii, D.; Levenspiel, O. Fluidization Engineering; John Wiley and Sons: New York, 1969; Chapter 7.
Received for review February 4, 2002. Revised manuscript received December 2, 2002. Accepted December 4, 2002. ES020020W
