Incinerator Design and Operation: The Robust Approach to PCDD/F Minimization

William F. Carroll, Jr., Ph.D.

April 2003


Thousands of experiments have been conducted over the past twenty-five years exploring the relationship between combustion conditions--including the amount of chloride in incinerator feed--and the generation and emission of dioxins and furans (PCDD/F). Two sources of chloride have been studied in great detail: salt and polyvinyl chloride (PVC) plastic. The data are diverse because there is a virtually infinite combination of fuels, combustion conditions, and designs of experiments ranging from one gram samples burned in a quartz tube1, to full-scale experiments where PVC and salt are purposely added and removed2,3.

Recently, Pat Costner of Greenpeace published a report on the Greenpeace website entitled "Chlorine, Combustion and Dioxins: Does Reducing Chlorine in Wastes Decrease Dioxin Formation in Waste Incinerators?"4 In it, she asserts: 1) Small-scale combustion is well-controlled and reproducible and clearly shows a correlation between the chloride content of waste and dioxins generated and emitted from combustion. 2) Data from experiments performed in large-scale combustors is prone to experimental error and is irreproducible; as a result, no clear conclusions can be drawn about the relationship between chloride and dioxin in those devices. 3) Despite the equivocal nature of data from large-scale combustors, small-scale results should be considered indicative of the overall mechanism of combustion and a material policy (elimination of PVC) should be put into place. She claims such a policy would result in reduction or elimination of PCDD/PCDF from combustion emissions.

In response to literature cited and analysis presented in Costner4, this paper makes four points: 1) Small- and intermediate-scale combustors do not provide clear evidence that chloride concentration is the controlling influence on PCDD/F generation. 2) Full-scale combustors, because they represent real-world operation and not a simulation, provide the best information. 3) The known catalytic formation of PCDD/F in combustion gases argues against a strong impact of the amount of chloride present in the fuel. 4) The best policy option for incinerators is to utilize combustion and emission technology that is robust with respect to any reasonable fuel.

Chloride and the PCDD/F Molecule It is literally true that in order to make chlorinated dioxins and furans-or chlorinated anything, for that matter-the chemical element chlorine is required. On the other hand, there is a more than a million-fold oversupply of chloride already present in normal wastei above what is needed to make the PCDD/F emitted. In fact, it would be impossible to "starve" the chloride out of the feed, since even ambient air alone can contain nearly 20 times as much chlorideii as is necessary to make that same PCDD/F5. Given these facts, it is difficult to imagine how micromanaging the chloride or PVC in the waste could have a major effect on PCDD/F generation.

1) Small- and Intermediate-scale Combustion: Scale Matters. Many experiments but results are not consistent.

Laboratory experiments In general, small-scale experiments are easier to conduct and to control than experiments conducted in full-scale apparatus. However, experimental accessibility and diverse experimental conditions and results provide important evidence of the complexity of the PCDD/F formation mechanism and the importance of design and operation. Costner cites the following articles:

    Wikstrom, et al.,19966 This experiment uses a pelletized, artificial fuel containing cellulosics, known catalytic metals, and polyethylene, varying chloride as a salt or PVC from 0-2%. They find no difference in effect between salt and PVC, and no effect of chloride amount below a threshold of about 1%.

    This laboratory has expanded upon these results in later experiments7. In testing the effects of good, intermediate and poor combustion in addition to fuel composition, these authors find:

      ".the most important variable for changes in the PCDDs/Fs and PCBs formation was disturbance in the combustion condition and not the variation in chloride content of the fuel. Furthermore, no differences in formation between the chloride sources could be seen."

    Four of the seventeen PCDD/F species measured in the article correlate with chloride content. By far, the majority do not, and the overall conclusion of the authors is that chloride is not a leading variable.

    Hatanaka, et al., 20008 Hatanaka tests artificial wastes containing salt, PVC and copper chloride as a catalyst for PCDD/F formation. The types and amounts of PCDD/F produced in this experiment differ from those of a normal combustor, and the authors note this, calling into question the applicability of this experiment to full-scale apparatus.

    PCDD/F also correlates strongly with carbon monoxide (CO), and increasing CO concentration is usually indicative of poorer combustion. "As Cl content of the waste was increased, CO concentration in the flue gas became higher and more PCDD/Fs were formed."

These experiments are simply telling us that combustion conditions are changing as the feedstock is changing; thus, combustion conditions may, in fact, be the most important variable. At the very least, the two potential causes are indistinguishable. This experiment was not set up to maintain good combustion regardless of fuel as is the case for most full-scale combustors. If it had been the results may well have differed.

Hatanaka, et al. added to their interpretation of the work in a later publication not cited in Costner9.

    ".CO concentration in the gas became higher and more PCDD/Fs were formed as chlorine content in waste increased. This means that combustion conditions indicated by CO are strongly related to the PCDD/Fs formation during incineration."

Later experiments were conducted at a different combustion temperature, holding the fuel (and Cl concentration) constant. The researchers reported:

    ".the PCDD/Fs concentration at 700C is significantly lower than at 900C in the experiment of the waste with the same Cl content."

Thus, a change in combustion temperature had a significant effect on PCDD/F in the absence of change in chloride concentration in the fuel.

Gullett, et al., 200010 This is the least applicable study to full-scale combustion. Gullett's experiments were conducted as uncontrolled combustion in an open industrial metal drum. This condition is diametrically opposed to the controlled, closed operation of a full-scale combustor.

While experiments run with 7.5% added PVC yielded the highest PCDD/F TEQ values, these experiments also demonstrated poor combustion. Statistical models constructed by the authors show that the most important variables relate to combustion rather than chloride. Gullett reports in a later article:11

    "The most significant one-predictor model actually contains an interactive term, CO*TC6, which is the product of CO emissions and the temperature at the uppermost portion of the barrel..The logarithm of Cu emissions or waste Cl content is also significant (based on its P<0.05), but not as effective at predicting emissions as the aforementioned models."

As another indicator of the importance of good combustion, when wet waste of the control composition (0.2% PVC) was burned, dioxin generation increased five-fold beyond that generated from waste containing 1% added PVC, although these two results were statistically indistinguishable.

2) Full-scale Incineration: Data for full-scale incineration exists and should be used. Scale-up, design and operation are the most important variables.

Hundreds of full-scale incinerator tests have been conducted, and robust statistical methods-the most reliable approach--have been used to interpret the data. In its landmark study of data from 170 combustors, the American Society of Mechanical Engineers states:

    "While some laboratory experiments show there is a functional relationship between chlorine input and PCDD/F concentrations in the products of combustion under certain conditions, the effect is much smaller than the effect of confounders like combustor design, operating practices and the normal variability found in emission measurements made at commercial scale systems. Whatever effect chlorine has on stack gas PCDD/F concentrations from waste combustors is masked by these other variables12."

This study, reviewed by over 25 experts in the field of combustion, found that for about 10 percent of commercial combustors, PCDD/F increases significantly with increasing chloride; for about 10 percent, PCDD/F decreases with increasing chloride, and for about 80 percent there is no significant effect observed. Yet, the ASME researchers are not the only ones who believe that changes in chloride concentration have a relatively minor effect, if any, on full-scale combustors.

Gullett13, et al note:

    "There is poor correlation between total chlorine in waste streams and formation of polychlorinated dibenzodioxin and polychlorinated dibenzofuran (PCDD/F) during waste combustion."

Everaert and Bayens14 carried out a similar review of large-scale combustors in order to determine mechanism of formation. They found positive correlation between electrostatic precipitator (ESP) temperature and PCDD/F. When values were normalized for ESP temperature:

    "The scatter and/or dependency of results disappears completely and PCDD/F levels achieve nearly constant values. It might therefore be concluded that the main correlating parameter for the PCDD/F emission level is the temperature of the ESP."

The authors also report "no correlation of PCDD/F with water, oxygen, hydrogen chloride (HCl), sulfur dioxide (SO2) and polycyclic aromatic hydrocarbons (PAHs)"

Accurate transcription of the EPA document15 cited by Costner agrees, stating that small-scale incineration is at best a poor indicator of full-scale performance. Section 2.4.2, "Review of Full Scale Combustion Systems" begins:

    "The review of experimental data clearly indicates an association between chlorine content of feed/fuels and potential synthesis of CDDs and CDFs." ("potential" omitted in Costner; emphasis added.)

So as not to be misunderstood, however, the document continues immediately following:

"Paradoxically, the review of full-scale operating incineration processes does not yield such unequivocal results indicating that complex kinetic events make strong associations difficult in full-scale systems."

EPA's opinion is clear: there are differences in chemistry between large- and small-scale operations. In the section "Potential Prevention of CDD/CDF Formation in Combustion Systems," the EPA authors cite experiments adding sulfur compounds or calcium oxide to inhibit catalytic formation of PCDD/Fs. While EPA is very well aware of chloride reduction as a "potential" solution, no mention is made of removing chloride or PVC. Moreover, in the "Integrated Summary and Risk Characterization" of the Dioxin Reassessment, the question is answered directly:

    "Although chlorine is an essential component for the formation of CDD/CDFs in combustion systems, the empirical evidence indicates that for commercial scale incinerators, chlorine levels in feed are not the dominant controlling factor for rates of CDD/CDF stack emissions. Important factors which can affect the rate of CDD/CDF formation include the overall combustion efficiency, post-combustion flue gas temperatures and residence times, and the availability of surface catalytic sites to support CDD/CDF synthesis"

3) Dioxin Formation Mechanisms: PCDD/Fs are formed under the influence of catalytic metals and largely at well-known temperatures. Fly ash surfaces are also involved.

Much of the work done on PCDD/F and combustion in the past twenty-five years has been oriented toward understanding the mechanism of formation. The sheer number of studies stands in testament to the complexity of this question; still, much is known and has been translated from the laboratory to operational technology.

    a) Formation temperature: The temperature for maximum formation of PCDD/F is between about 250 and 400 degrees C16,17,18. Combustion gas systems such as "hot-side electrostatic precipitators" hold combustion gases at these temperatures and will tend to be incubators for PCDD/F19. On the other hand, so-called "fast-quench" systems that cool the gases quickly through these temperatures discriminate against formation.

    b) Surfaces and metals: It is well-known that active fly ash containing elemental carbon as soot or carbon black can yield PCDD/F when simulated combustion gases are passed over it 14,20,21. This seems to be due to the action of small amounts of metal chlorides-usually copper-on the carbon black attached to the fly ash. The metal chlorinates the carbon, and eventually, aromatic structures detach from the carbon particle. These can act as precursors to PCDD/F or be PCDD/F themselves.

    c) Further chlorination and dechlorination: Experiments have demonstrated the possibility of active exchange of chlorine atoms on and off the basic skeletons of dioxins and furans. This exchange helps to determine the mix of species found in the final emissions22.

It appears not to be true, however, that simply increasing the amount of chloride in the combustion gas impacts these mechanisms. Catalyst is the scarcest material present in these incineration systems, and there is already a huge excess of chloride.

Combustors fitted with apparatus designed with knowledge of these mechanisms perform significantly better than those of earlier design. US EPA estimates that implementation of its Maximum Achievable Control Technology (MACT) will be responsible for reducing PCDD/F generation from municipal and medical waste combustors from about 1700 g TEQ to about 20 g TEQ in the period 1995-200423. This 99% reduction comes without modifying the material entering the incinerator; e.g., no restriction on chloride.

4) Incineration Policy: The best control option for incinerators is combustion and emission technology that is robust in the face of any reasonable fuel.

If there is any general conclusion to be drawn from twenty-five years of research into dioxin generation in combustion, it is that combustors vary. Similarly, emissions from combustors vary, even with respect to dioxin emissions and correlation with chloride. Carefully chosen literature can demonstrate that there is a positive correlation between input chloride and PCDD/F (e.g, more chloride = more dioxin); similar care can find experiments showing negative correlation (e.g., more chloride = less dioxin). Conversely, if the correlation is correct, reducing chloride from the input to these two sets of combustors would yield two different results: less dioxin emission from the first set, but more dioxin emission from the second. Adjusting the chloride content of the combustor fuel is clearly not a robust PCDD/F reduction policy.

Table 1: PCDD/F by effluent, before and after application of technology
Before Technology
(pct of total)
(pct of total)
Flue gas 40.1 (65) 0.006 (<1)
Fly ash/Treatment solids 21.1 (34) 1.4 (83)
Bottom Ash 0.3(<1) 0.3 (16)
Total 61.6  

On the other hand, there is scientific knowledge regarding the temperature of formation and impact of metal catalyst on the reactions that lead to dioxin. This knowledge leads to technology that reduces both the generation and emission of PCDD/F--technology which has been specified by governments in Europe, Japan and the United States over the past decade. It is the application of this technology that has, and will, lead to the 99% reduction in dioxin emissions EPA foresees.

Similarly, it is patently not true that newer technology simply shifts PCDD/F from one medium to another. EPA's observation that a change in temperature of gases entering the ESP greatly reduces dioxin generation in that device is simply one example19. Less dioxin is made in modern systems, and less is emitted.

"Scrub" indicated use of a Ca(OH)2 semi-dry scrubber

A direct comparisons of "before" and "after" application of modern technology have been published24,25. In this paper, sequential application of technology shows the impact of different strategies (Figure 1). Reduction of greater than 99% in flue gas emissions is obtained without management of combustor fuel. Generation-and not just emission--of dioxin is decreased markedly as shown in Table 1.

Contrary to the conclusions of Costner's paper, the key to reduction of generation and emissions of PCDD/Fs is proper design and operation of the combustor. This means specification and operation of pollution control equipment and utilization of good combustion practices like the "Three T's": Time in the combustion zone, Temperature of combustion and combustion gases, and good Turbulence. Successful governmental policy reflects this and requires it as a condition for operation of combustors. Once this is done correctly little, if anything, is to be gained by other measures like removal of chloride.


(1) Vikelsoe, J.; Johansen, E. Chemosphere 2000, 40, 165-175.

(2) Visalli, J. R. J. Air Poll. Cont. Ass'n 1987, 37, 1451-1463.

(3) Pandompatam, B.; Kumar, Y.; Guo, I.; Liem, A. J. Chemosphere 1997, 34, 1065-1073.

(4) Costner, P. "Chlorine, Combustion and Dioxins: Does Reducing Chlorine in Wastes Decrease Dioxin Formation in Waste Incinerators?," Greenpeace, 2001.

(5) Grosjean, D. Environ. Sci. Tech. 1990, 24, 77-81.

(6) Wikstrom, E.; Lofvenius, G.; Rappe, C.; Marklund, S. Environ. Sci. Tech. 1996, 30, 1637-1644.

(7) Wikstrom, E.; Marklund, S. Chemosphere 2001, 43, 227-234.

(8) Hatanaka, T.; Imagawa, T.; Takeuchi, M. Environ. Sci. Tech. 2000, 34, 3920-3924.

(9) Hatanaka, T.; Imagawa, T.; Kitajima, A.; Takeuchi, M. Organohalogen Compd. 2001, 50, 430-433.

(10) Gullett, B. K.; Lemieux, P. M.; Winterrowd, C. K.; Winters, D. L. Organohalogen Compd. 2000, 46, 193-196.

(11) Gullett, B. K.; Lemieux, P. M.; Lutes, C. C.; Winters, D. L.; Winters, D. L. Chemosphere 2001, 43, 721-725.

(12) Rigo, H. G.; Chandler, A. J.; Lanier, W. S. "The Relationship Between Chlorine in Waste Streams and Dioxin Emissions from Waste Combustor Stacks," American Society of Mechanical Engineers, 1995.

(13) Gullett, B. K.; Sarofim, A. F.; Smith, K. A.; Procaccini, C. Trans IChemE 2000, 78B, 47-52.

(14) Everaert, K.; Baeyens, J. Chemosphere 2002, 46, 439-448.

(15) "Exposure and Health Reassessment of 2,3,7,8-Tetrachlorodibenzo-p-Dioxin (TCDD) and Related Compounds, Part I: Estimating Exposure to Dioxin-Like Compounds, Volume 2: Sources of Dioxin-Like Compounds in the United States.," US Environmental Protection Agency, 2000.

(16) Stieglitz, L.; Bautz, H.; Roth, W.; Zwick, G. Organohalogen Compd. 1995, 23, 319-322.

(17) Addink, R.; Altwicker, E. R. Organohalogen Compd. 1998, 36, 73-76.

(18) Wikstrom, E.; Persson, A.; Marklund, S. Organohalogen Compd. 1998, 36, 65-68.

(19) "The Inventory of Sources of Dioxin in the United States," U.S. Environmental Protection Agency, 1998.

(20) Huang, H.; Buekens, A. Sci. Tot. Env. 1996, 193, 121-141.

(21) Hatanaka, T.; Imagawa, T.; Takeuchi, M. Chemosphere 2002, 46, 393-399.

(22) Iino, F.; Tsuchiya, K.; Imagawa, T.; Gullett, B. K. Environ. Sci. Tech. 2001, 35, 3175-3181.

(23) "Database of Sources of Environmental Releases of Dioxin-like Compounds in the United States," US Environmental Protectiion Agency, Accessed December 17, 2001, 1998.

(24) Abad, E.; Adrados, A.; Caixach, J.; Rivera, J. Environ. Sci. Tech. 2002, 36, 92-99.

(25) Abad, E.; Caixach, J.; Rivera, J. Chemosphere 2003, 50, 1175-1182.

*The ASME study was partially funded by The Vinyl Institute and the Chlorine Chemistry Division of the American Chemistry Council, industry associations.
iAssume 10 ng TEQ/cubic meter of air emissions (dscm)-50-100 times the currently allowed emissions. This is ca. 400 ng total/dscm. At 6 chlorides average per D/F, this is approximately 1 nanomole PCDD/F per dscm or 6 nanomoles chloride/dscm. At 7000 m3 air/ton waste, 42 micromoles (1.5 mg) chloride is emitted as dioxin per ton of waste. Normal MSW contains approximately 0.5% chloride, so 5000 g chloride enters the combustor per ton of waste. This means that even under these poor combustion conditions there is well over 3,000,000 times as much chloride present in the waste as is required to make the PCDD/F.
ii Grosjean measures up to 3.9 micrograms/ m3 chloride in ambient air. At 7000 m3 air/ton waste, 27 mg chloride is introduced with the air, an 18-fold excess over what would be needed to synthesize PCDD/F.

home | Dioxin TRI Data - Cl2 Sector | sources & trends | dioxin & health | q & a | links | contact us

Copyright 1996 to Chlorine Chemistry Division of the American Chemistry Council. All Rights Reserved.
Terms and Conditions | Privacy Policy

Back to Index