This paper presents data on the source of PCDDs/PCDFs in cement plant stack emissions, demonstrating that these emissions are not necessarily all created by the process. The data potentially demonstrates that some of the PCDDs/PCDFs in the emissions are pre-existing dioxins and furans that are vaporized out of the kiln raw feed. This calls into question the assumption that PCDD/PCDF emissions can be decreased below a certain point. The EPA has acknowledged resuspended PCDD/PCDF emissions, such as from forest fires, as a 'reservoir source' of these emissions. Based on the data presented here, there may be a basis for an acknowledgement by the EPA of a site specific minimum level of PCDD/PCDF emissions attributable to the kiln feed raw materials as a 'reservoir source.'
There has always been the contention that PCDD/PCDF's are emitted predominately due to combustion of hydrocarbons. EPA's Boiler and Industrial Furnace (BIF) regulation, the Cement Kiln Dust (CKD) report, the Dioxin Reassessment and now the Maximum Achievable Control Technology (MACT) regulation all rely on this concept. The discussion of how dioxins are formed has been long and involved but with respect to cement kilns, no definitive set of mechanisms has been demonstrated to account for dioxin emissions. It is understood and generally accepted that there is a relationship between dioxin emissions and temperature. But what temperature ranges and at what location in the process is in some debate. Additionally, there is the belief that high hydrocarbon content in the kiln raw feed may produce higher dioxin emissions. For certain kilns this is true, for others it is not. EPA's proposed MACT standard for cement kilns that burn hazardous waste of 0.2 ng/dscm (TEQ) is based on this temperature formation scenario. The concept that dioxin emissions may be the result of resuspension of dioxins that already exist in nature, i.e. in the kiln feed, has been discounted by the EPA. The reason for this is simply that the EPA believes dioxin are man-made compounds only grudgingly and rarely admitting 'these compounds may have always been present in the environment'.1 As will be shown in this paper, EPA's understanding of cement kilns dioxin sources is grossly incomplete, which may have a profound impact on the implementation of the MACT standards.
Presentation of data
This paper presents PCDD/PCDF data from analyses of the emission and raw feed input of five cement kilns. Kiln 'A' is a large pre-heater precalciner kiln that utilizes hazardous waste fuel. This data was obtained during a series of BIF compliance tests. Kiln 'B' and 'E' are large long wet kilns that utilize hazardous waste fuel. This data was obtained during a series of BIF compliance tests. Kiln 'C' and Kiln 'D' are moderately sized long wet kilns that do not utilize hazardous waste fuel. The data was obtained during a series of voluntary tests to obtain cement kiln Hazardous Air Pollutant (HAP) data. Table 1 presents PCDD/PCDF emissions values and input values present in the raw kiln feed.
The PCDD/PCDF emission or raw feed input values are for the tetra through octa congeners. The stack emissions values are reported per EPA's reporting convention, that is 'non-detect' and 'EMPC' values are reported as zero's. For the purpose of this paper, the PCDD/PCDF values in raw feed are reported slightly differently. The total PCDD/PCDF values include the detection limit (DL) values for each non-detect (ND). This may overstate the input value for total PCDD/PCDF in raw feed. The column next to this '% as ND' indicates the weight percent of the total PCDD/PCDF that is represented by non-detects. A higher percentage is indicative of a raw feed with very little PCDD/PCDF. The raw feed TEQ values however, follow EPA reporting convention. The reason for reporting the DL values in the total PCDD/PCDF for the raw feed input is that even though these are very low concentrations, the input rate can be substantial. As an example for Kiln A, the raw feed input rate is 250+ tons per hour. In this manner, it is possible to present a worst case input rate, ('Total PCDD/PCDF'), and a possible diminishment from that value, ('% as ND'), as well as the well known 'TEQ' value.
The data for Kiln A also allows additional insight to PCDD/PCDF input from the raw feed. In preheater/precalciner kilns, process gases are used to dry the raw feed during milling. Consequently, any hydrocarbons in this process gas may be scrubbed out and deposited onto the raw feed possibly giving a false PCDD/PCDF input value. This raw feed would then enter the process system where some of these hydrocarbons would be stripped off to be redeposited onto the raw feed. In the fall of 1995, Kiln A conducted two PCDD/PCDF stack emission tests. One with the process gas passing through the mill and one with the process gas by-passing the mill. These emission values are presented in Table 2.
Figure 1 is a block diagram of the Kiln A process. With the known PCDD/PCDF rates in the appropriate places. The figure easily demonstrates that stream number 3, (the PCDD/PCDF feed rate to the kiln), is made up of stream number 4, (the PCDD/PCDF feed rate from the raw feed constituents), and the difference between the PCDD/PCDF content in stream number 2, (the emission when the mill is not operating), and stream number 1, (the emission when the mill is operating). Consequently, stream number 4 is equal to 1.57E-07 g/sec minus (2.9E-9 g/sec - 6.2E-10 g/sec) or 1.55E-07 g/sec.
Essentially, 98+% of the PCDD/PCDF that enters the kiln process must come from the raw feed constituents, primarily quarried limestone. In this particular kiln process, the waste dust is dust removed from the kiln bypass duct. As such, this dust does not have measurable quantities of PCDD/PCDF, nor does the clinker. This leads to the conclusion that approximately 98% of the PCDD/PCDF that enters with the raw feed is destroyed in the kiln process with another 1.4% removed from the gases by the mill to be recycled, 98% of it subsequently destroyed.
Discussion of the data
The data from Kiln A very clearly indicates that substantial amounts of PCDD/PCDFs are entering the kiln system from the raw feed constituents, essentially two to three times the order of magnitude as is emitted. Kiln A is very efficient in destroying nearly all of this input. If this were not the case, Kiln A would not achieve the MACT standard of 0.2 ng/dscm.
Kiln B, however, is less fortunate in that it produces additional PCDD/PCDF. As noted in the introduction, it is uncertain why this kiln produces dioxins when similar long wet kilns do not. At the designated 0.2 ng/dscm standard, the allowable emission rate is 12x10-9 g/sec (TEQ). In Kiln B's case, up to half of that quantity enters with the raw feed.
Kiln C also produces dioxins. This kiln does not utilize hazardous waste fuel and is not subject to the MACT standard. However, under the MACT standard a mere 1% of the allowable PCDD/PCDF emissions would enter with the raw feed.
Kiln D does not appear to produce significant PCDD/PCDFs, nor were any detected in the raw feed. Calculating backward from PCDD/PCDF emissions to a PCDD/PCDF concentration in the raw feed would indicate a PCDD/PCDF (TEQ) concentration in the raw feed in the range of 4x10-15 g/g. This is well below the detection limit of 0.1 ppt (1.0E-13) typical in raw feed analysis. Conceivably then, the PCDD/PCDFs in the raw feed are the source of the PCDD/PCDF emissions for this kiln.
Kiln E also produces PCDD/PCDFs. However, the 7.4E-09 g/sec (TEQ) input in the raw feed, if all of it were stripped out during processing, would result in an emission concentration of 0.12 ng/dscm, more than half of the allowable emission under the MACT standard.
Clearly, there are PCDD/PCDFs entering kilns with the raw feed and the input rate of these PCDD/PCDFs are highly variable from facility to facility. This input ranges from orders of magnitude more PCDD/PCDF than would be allowable as emission under the MACT standard, to virtually immeasurable small quantities. These PCDD/PCDFs are in the raw feed constituents; possibly in the limestone, shale, clay, sand or iron bearing materials and in the case of the wet process kilns in the slurry water. Based on the differences in congener profile, it is also possible that PCDD/PCDFs with low toxic equivalency factors (TEFs), such as OCDD and OCDF which are the most common dioxins and furans found in the raw feed, are being converted to more toxic forms. In any case, concentrations of PCDD/PCDF (TEQ) in the range of 0.4 ppt in dry raw feed and 0.2 ppt in slurry feeds may cause the kiln's emissions to exceed the 0.2 ng/dscm emission limit set forth in MACT.3 At the present, these values are very near the detection limit for PCDD/PCDF analysis of the raw feed matrix. For the water used to slurry the wet process kiln feed, the analytical detection limit is in the 1 to 2 ppq (1 to 2E-15) although based on preliminary and currently ongoing research some river water may have PCDD/PCDF levels as high as 1 ppt, virtually certain to cause the PCDD/PCDF input to the kiln to exceed the 0.2 ppt noted above.
Based on informal discussions with EPA representatives, the impact of this data on the implementation of the MACT standard for kilns that utilize hazardous waste is potentially profound4. The reason for this is that MACT specifically exempts from emissions controls those contaminants that are in raw materials. As this paper has shown, there are significant and variable quantities of naturally occurring PCDD/PCDFs entering cement kilns with the raw feed. Consequently the EPA may well have to make some accommodation for these naturally occurring PCDD/PCDFs when establishing PCDD/PCDF emission limits for specific cement kilns.