David Gossman
President, Gossman Consulting, Inc.

Myron Black
Manager, Environmental Affairs, Holnam Cement Co.

Mark Ward
Senior Process Engineer, Holnam Cement Co.

 

Presented at the AWMA International Specialty Conference on Waste Combustion in Boilers and Industrial Furnaces April, 1990
 

This study was taken to develop information on the fate of trace metals introduced with hazardous waste fuel and other raw materials into wet portland cement kilns. Such information is necessary to evaluate what changes in emissions can be expected and whether an environmental impact can be expected that makes regulatory controls appropriate. The study seeks to determine the potential for emissons on a metal-by-metal basis by determining the fate of metals entering the process under both baseline conditions and while burning hazardous waste fuels. Particular attention is paid to metal fate based on the point where the metal enters the process, i.e. with the raw feed or with the fuel. Three kilns at two plants were analyzed by taking samples of all material inputs and outputs including raw feed, fuel, clinker and kiln dust. Multiple sets of data were taken both under baseline conditions and while the kilns were using various quantities and qualities of hazardous waste fuel. Test results indicate that although relative volatility is an important factor in determining metal fate, other processes such as the development of recirculating loads may impact the degree of incorporation of the metals in clinker. It was also shown that with a few exceptions that most trace metals are present in the kiln in larger quantities from the raw materials and coal used to manufacture cement than are introduced with hazardous waste fuel. Metals tested were antimony, arsenic, barium, beryllium, cadmium, lead, mercury, nickel, selenium, silver, thallium, vanadium and zinc. EP Toxicity and TCLP studies of the cement kiln dust were also performed.

Introduction

In light of rising concerns by state and federal regulatory officials over the potential for dangerous metal emissions from cement kilns using hazardous waste fuels, this study was undertaken. EPA has financed metal fate studies on hazardous waste incinerators but has not performed similar testing on cement kilns.

This study seeks to determine the potential for emissons on a metal-by-metal basis by determining the fate of metals entering the process under both baseline conditions and while burning hazardous waste fuels in wet process cement kilns. Particular attention is paid to metal fate based on the point where the metal enters the process, i.e. with the raw feed or with the fuel. EP Toxicity and TCLP studies of cement kiln dust were also performed.

Methodology

Table I lists the metals tested for in various input and output materials of the cement manufacturing process along with EP Toxicity/TCLP tests on the cement kiln dust. This data was accumulated using the sampling protocol listed. A total of fourteen complete sets of samples and tests have been concluded; three each on kilns 1 and 2 during baseline conditions, one while burning hazardous waste fuel in kiln 1, four while burning HWF in kiln 2, and three while burning HWF in kiln 3.

Initial laboratory analysis of the samples using EPA SW-846 digestion method 3050 produced data that was obviously low and wrong. Checks with the laboratory confirmed that the samples did not fully dissolve during digestion. A search for a better digestion methodology was then undertaken. Table II contains a method obtained from the National Bureau of Standards that was eventually utilized. This method fully dissolved all samples but the coal sample. The coal sample appears to have had all of its mineral and a portion of its hydrocarbon content dissolved leaving a carbonaceous residue from which the metals appear to have been fully extracted. Analysis of the data reveals that the clinker samples were contaminated with Cr and Ni from the grinding equipment used prior to digestion. This date was therefore not used in the analysis. All other samples appear to have valid Cr and Ni data. Mercury (Hg) analysis was performed on samples using EPA SW-846 Method 7470-7471. Quantification of each metal was performed using GF-AA, ICP-AE, or ICP-MS depending on potential interferences and detection limits for the selected metals.

Results

Appendix 1 contain the metal data for the laboratory along with spreadsheet calculations on a metal-by-metal basis. None of the cement kiln dust samples showed detectable levels of EP Toxic/TCLP metals.

Antimony (SB). Only during two tests did levels exceed 1 ppm in kiln feed. Only during one test did levels exceed 1 ppm in coal. Levels in the hazardous waste fuel ranged from 6.5 to 49 ppm. This is consistent with anticipated levels of up to 100 ppm routinely found in blended hazardous waste fuels. Only during test on kiln 2 was there a statistically significant increase in antimony concentrations in kiln dust when compared to antimony levels in clinker. There was a statistically significant tendency for the antimony to concentrate in the clinker rather than the dust in kiln 3. This was likely because a large portion of the antimony entering the system was in the fuel. An average of 95% of the antimony exited the system in the clinker in kiln 3.

Arsenic. Arsenic concentrations in the kiln feed ranged from <1 to 20 ppm and <1 to 14 ppm in the coal. The hazardous waste fuel had <1 to 8.8 ppm arsenic, well within the 200 ppm maximum expected. Because the background amounts of arsenic entering the process with the kiln feed and coal were consistently over 100 times the amounts found in the HWF, no correlation between arsenic amounts in the HWF and kiln dust was seen. The data suggests that the arsenic entering with the raw material is concentrated to an extent in the dust and then the remainder becomes part of the clinker. Arsenic may form a number of different thermally stable compounds, an example of which is potassium arsenate which melts at 1310C. More likely the arsenic becomes entrained within the crystal structure or lattice of the portland cement clinker compounds. This sort of chemical entrainment or fixation may be why significant amounts of arsenic remain in the clinker.

Barium (Ba). Barium concentrations in the kiln feed ranged from a very consistent 35 to 82 ppm and from 15 to 540 ppm in coal. The hazardous waste fuel had 110 to 610 ppm, which is consistent with typical barium levels in HWF. Analysis of the data indicates that most of the barium in the dust comes from the raw feed as it enters the kiln without any concentrating effect. This is consistent with the well documented tendency for barium to readily substitute for calcium in the compounds formed in portland cement clinker.

Beryllium (Be). Beryllium concentrations in the kiln feed ranged from <0.5 ppm to 1.4 ppm and <0.5 to 1.8 ppm in the coal. No beryllium was detected in HWF at the 1 ppm detection limit. The only statistically significant tendency was for beryllium to concentrate itself somewhat in the clinker. This is consistent with the high boiling point of Be of 3900C.

Cadmium (Cd). Kiln feed, clinker and coal showed no detectable levels of cadmium. Likewise, kiln dust showed no detectable levels under baseline conditions. HWF contained 6.5 to 71 ppm cadmium, well within the expected range of 0-500 ppm. Figure 1 clearly shows the relationship between the amount of cadmium in the dust and the amount in the HWF. The data indicates that cadmium volatilizes and exits the kiln in the dust.

Chromium (Cr). Chromium levels in the kiln feed ranged from 7.1 to 66 ppm. Concentrations in coal were <1 to 43 ppm. HWF contained 60 to 280 ppm. Levels of up to 2,000-3,000 ppm chromium are found in waste fuel. There was no statistically significant correlation between the total input of chromium and amounts leaving the system in the kiln dust. The only statistically significant relationship found in the data was that levels of chromium in the dust were most of the time lower than that in the dry raw feed. This indicates a strong tendency for chromium to concentrate in the clinker as might be expected given a boiling point of 2762C. This data provides no reason to believe that chromium entering the process with the HWF will be emitted from the stack. The only apparent source of chromium in the dust is from the raw feed which provided substantially larger quantities of chromium to the system than any other source.

Lead (Pb). Concentrations of lead in the kiln feed ranged from 1 to 15 ppm and 2 to 25 ppm in the coal. The HWF had 350 to 1800 ppm Pb. Figure 2 illustrates the strong correlation between the amount of lead from the fuel and the amount exiting with the kiln dust. Approximately 71% of the Pb entering the system with the fuel exits with the kiln dust. Clinker lead concentrations varied from <1 ppm to 160 ppm. This indicates a significant recirculating load of lead in the hotter zones of the kiln that periodically flushes out with the clinker. This is a familiar pattern in cement chemistry and occurs with a number of low concentration clinker constituents. Nevertheless, these tests indicate a strong tendency for Pb to concentrate in the kiln dust.

Mercury (Hg). Mercury was only detected above 0.2 ppm in four of the samples tested. It was found at 0.4 ppm in three samples of HWF and at 0.26 ppm in a dust sample. This is consistent with the experience at other plants, indicating that mercury is generally quite low in cement plant raw materials as well as HWF.

Nickel (Ni). The concentration of nickel in the kiln feed ranged from 3.2 to 28 ppm and 20 to 170 ppm in the coal. HWF concentrations of nickel ranged from 8.9 to 35 ppm, less than the coal being replaced.

Selenium (Se). The selenium concentrating in the raw feed ranged from <0.5 to 31 ppm and <0.5 to 28 ppm in the coal. The HWF had <0.5 to 33 ppm of selenium. Because of the relatively large quantities of selenium present in the raw feed, in most cases, no discernible effect on amounts in the kiln dust based on quantities in the fuel could be determined. The relationship between total selenium input and the amounts in kiln dust seemed to behave similarly to lead. In addition, as with lead, the amount input is on many occasions larger than the amount output, indicating the development of a recirculating load.

Silver (Ag). Silver was not detected in the raw feed or in the coal down to <3 ppm. Trace levels of silver from <1 to 3 ppm were found in the HWF. Similarly, trace levels of silver were found in the clinker and kiln dust at levels of 1 to 3 ppm. This limited data did indicate with statistical significance that the silver tended to concentrate in the clinker rather than the kiln dust. This was expected based on silver's boiling point of 2212C.

Thallium (Tl). Thallium was not detected to <1 ppm in kiln feed or coal. Only one HWF sample showed a detectable level of thallium at 1.2 ppm. The only detectable levels of thallium found was in the kiln dust at concentrations of 0.59 to 3.7 ppm. Given the somewhat lower boiling point of thallium compared, for example, to lead, this was expected. Any thallium in the system can be expected to concentrate in the dust.

Vanadium (V). Vanadium concentrations in kiln feed ranged from 9.6 to 48 ppm and from 130 to 640 ppm in coal. HWF had 3.4 to 18 ppm vanadium. The data does indicate that some of the vanadium from the fuel find its way to the kiln dust. Essentially, vanadium is evenly distributed in the process output materials, clinker and kiln dust although a slight concentrating effect in the clinker may be possible.

Zinc (Zn). The concentration of zinc in kiln feed ranged from 11 to 30 ppm and 10 to 130 ppm in coal. HWF had from 500 to 1300 ppm zinc as expected. The data clearly indicates that overall zinc tends to concentrate in the dust as opposed to the clinker. However, a linear regression analysis indicates that clinker concentrations of zinc are related to HWF zinc level. The same analysis indicates that the concentration of zinc in the dust varies largely independent of any zinc input parameter. The consistently high input to output ration probable indicates the development of a recirculating load. Multiple regression analysis indicates that the percentage of zinc leaving the system in the kiln dust is inversely proportioned to percentage entering with the fuel when the dust wastage rate is used as the other independent variable. These are clear indications that where zinc enters the system has a strong influence on where it will exit.

Conclusion

Metal fate testing provides a valuable resource in determining the extent of stack testing required to protect health, safety and the environment. It can also be used to evaluate the potential impact or lack of impact that using HWF or any other raw material might have on cement clinker and kiln dust. Based on this experience, the choice for sample digestion method chosen can severely affect the results by as much 1-2 orders of magnitude.

Table III provides conclusions which classify the fate of metals in these wet process cement kilns. Arsenic, cadmium, lead, mercury, selenium and thallium will concentrate in the kiln dust although arsenic, selenium and lead will also exit in clinker. Beryllium, chromium, silver and possibly vanadium tent to concentrate in the clinker. Barium, nickel and vanadium are found in similar concentrations in both clinker and kiln dust, not concentrating in either. Antimony, beryllium, zinc and possibly arsenic, barium, chromium, nickel and selenium show indications that their route out of the kiln is dependent to some extent on where they came in. Finally, lead, selenium and zinc all showed signs of developing recirculating loads. Antimony and arsenic may also develop recirculating loads.