Back to Publications

Gossman Consulting, Inc.


Dr. Michael von Seebach

Senior Vice President, Southdown, Inc., Houston, Texas

David Gossman

President, Gossman Consulting, Inc., Hampshire, Illinois

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

Cement kilns have all the ingredients required for complete scrubbing of acid gasses. These ingredients are a broad range of gas temperatures from 220° to 3,000°F; gas retention times of approximately 30 seconds, high levels of turbulence, high concentrations of alkaline solids, including sodium and potassium oxides; and, in high concentrations, freshly created CaO. Despite this environment, HCl emissions are reportedly measured from cement kilns when EPA Method 26, as presently proposed, is used.

This paper describes how the analyses of the impinger solutions can be taken a step further. This is performed by testing the impinger train not only for chloride but also for a variety of cation concentrations. Specifically, the cations ammonium, sodium, potassium, and calcium were analyzed. The results clearly demonstrate the Method 26, as presently proposed, provides faulty results. The data proves conclusively that the chloride measured with Method 26 when applied to cement kilns originates from ammonium chloride and other chloride salts, and does not originate from HCl. This was further confirmed by analyzing the condensate at the transition between the hot and cold connectors between the heated filter box (250°F) and the ice­cold impinger bath. The analysis of the condensate clearly revealed that NH4Cl had condensated at temperatures below 220°F.

This validates previous proposals that HCl emissions from cement kilns are impossible. The results further indicate that EPA Method 26 as presently proposed is not applicable for cement kilns. As amendment, the impinger solutions should not only be analyzed for chloride but additionally using ion chromatography, for NH4+, Fe2+, Fe3+, Al3+, Ba2+, Ca2+, K+, Mg2+ and Na+. The presence of any of these cations would indicate the penetration of the filter by volatile or pseudo­particulate solids.

Introduction and Background

Southdown, Inc. is the parent company of Southwestern Portland Cement Company which owns and operates eight active cement plants within the continental U.S. Southdown is a U.S. owned company and most of its 11 kilns are already covered by New Source Performance Standards. Southdown, Inc. and Southwestern Portland Cement Company's primary business is the manufacturing of high quality Portland Cements. Some of Southwestern's cement kilns are currently utilizing hazardous liquid and solid waste in the beneficial reuse of these wastes through recycling, utilizing the waste as fuels to replace, partially, fossil fuels in cement kilns.

The very high acid scrubbing capabilities of cement and lime kilns (not aggregate kilns) is a well known phenomenon1. The ability of these kilns to scrub acid gases is based on the high turbulence and residence times of the gases loaded with highly alkaline, fine and reactive solids. These reactive solids include freshly decarbonized lime (CaO), in the process of formation. This provides for the complete reaction of any acid gases which might form in the combustion zone with the highly alkaline material tumbling the entire length of the kiln and in the suspension preheater (for the energy efficient type of system). Even the particulate emissions are highly alkaline, and maintain the alkalinity over an extended period. This clearly indicates that any chloride emissions are in the form of salts, not HCl.

Despite these facts, literature references2,3,4,5,6,7 and EPA supported stack tests continue to report HCl emissions from cement kilns. This strongly suggests the conclusion that some interfering species penetrates the sampling train filter producing erroneous "HCl" results. Indeed section 1.3 of proposed Method 26 states "Volatile materials which produce chloride ions upon dissolution during sampling are obvious interferences."8

Southdown completed stack tests at three cement plants where EPA proposed methodology for "HCl" testing was employed. To clear up this discrepancy in addition to testing the impinger trains used in these tests for chloride, a variety of cation concentrations were also determined.

Test Procedures and Results

Stack tests were performed by the suggested Method 26 by means of a gas pump. The particulate and gas samples were drawn out of the stack through an insulated probe. Gas velocity into the probe was controlled by the usual Pitoty tube arrangement. From the insulated tube through the stack wall, the gas sample carrying the particulates was brought into a heated area. In this heated area the filter arrangement holding back the major particulates was located. From there the gas sample was pumped through the impingers. Also the evaluation of the results of the stack tests were performed in accordance with the proposed Method 26.

The stack test results are shown in detail in Table I. The second column of this table shows the chloride content of the front and back impinger solutions. These data demonstrate that in all impinger solutions, with very few exceptions, Cl­­anions were found. In addition to the anion analysis, cations were also measured in the impinger solutions. Specifically, the cations NH4+ Na+ and K+ were determined and found in considerable quantities. Specifically, ammonium was found in large quantities. It was found that of 27 impingers tested all but three showed more than enough ammonium to account for all of the chloride as ammonium chloride. These data are shown in detail in Table I. In addition to the ammonium chloride found, there is enough potassium and/or sodium to account for the remaining chloride in two of those three cases in which insufficient ammonium was present to account for all chloride.

These results are graphically demonstrated in Figures 1 ­ 3. These figures show the measured chloride as a percent of cation carrying capacity for all measured plants in Florida, Kentucky and in Colorado. All figures show clearly that there are generally far more cations in the impinger solutions available than could be satisfied with chloride. Consequently the results clearly demonstrate that there must be chlorides and/or other pseudo particulate alkaline salts that penetrate through the filter and end up in the impingers.

In Table I, columns 2, 3, and 4 primarily suggest that ammonium chloride has penetrated through the filter into the impinger solution. Therefore a linear regression analysis of the data given in Table I columns 2 ­ 3 was performed. The results of this statistical correlation of Cl­ and NH4+ in the impingers is shown in Figure 4. This figure shows the chloride concentration in the impinger plotted versus the ammonium concentration. Only the front impinger was used for this correlation. The graph shows that there is a statistically significant correlation between the ammonium concentrations and the chloride concentrations in the front impinger. Unfortunately, there were an inadequate number of data points available from the back impingers to draw a statistically significant conclusion. However, the statistical analysis of the chloride and ammonium concentration in the front impinger clearly shows that the chloride primarily results from ammonium chloride penetrating through the filter.

This finding is supported by separate tests9 performed at Southdown's plant at Victorville, California. During stack testing in this plant a white condensate was found in the glass tube connector between the front half filter and the back half impingers in a Method 5 sampling train modified for HCl determination. Using aqueous dissolution followed by specific iron electrode analysis nearly stoichiometric amounts of ammonium and chloride ions were found. This result further confirms the nature of chloride emissions as ammonium chloride from cement kilns.


Since considerable amounts of cations, primarily ammonium, are found in the impinger solutions Method 26 as presently suggested cannot be used for measuring HCl emissions from cement kilns. In fact, the data presented in this paper clearly show that chloride is present as ammonium chloride or other chloride salts in the particulate emission of cement kilns.

The data further show the value of routinely monitoring cation concentrations when performing stack tests from alkaline sources using Method 26.

Recommendation of Modification to Method 26

If HCl testing using proposed Method 26 is absolutely necessary even in the face of data clearly demonstrating the unlikelihood of HCl emissions, the use of the following minor modifications is recommended:

1. Eliminate the third and fourth (caustic) impingers. This portion of the system is designed to capture Cl2 which is not an emission issue for cement kilns.

2. In addition to testing the impinger solution for chloride anions, test the solution using ion chromatography for the following cations: NH4+, Fe2+, Fe3+, Al3+, Ca2+, Ba 2+, K+, Mg2+, and Na+. The presence of any of these cations would indicate the likely penetration of the filter by volatile or fine particulate salts. The concentrations of these cations can be used to correct the Cl­ concentration by the appropriate stochiometric quantities.

3. Since this is an invalidated methodology for cement kilns, blank runs using the same sampling train and a high purity mix of CO2 and nitrogen should be performed. Values above the detection limits for both Cl­ and the cations should be used as a blank correction of the measured value.

4. Three runs of two hours duration should be performed as well as three runs of similar durations and flow rates for the blank determinations. Results should be averaged and the determination should include a minimum and maximum value for the 95% confidence range. This determination should consider both sampling and analytical errors.

The results of any such testing should be considered tentative until the EPA validates the method for use on cement kilns.


1. S. Sprung, "Technological Problems in Pyropressing of Cement clinker, VDZ Publications, Volume 43, Dusseldorf, 1982

2. M. R. Branscome et. al. "Evaluation of waste combustion in cement kilns at General Portland, Inc., Paulding, Ohio," Research Triangle Institute, draft report on EPA Contract 68­02­3149, Work Assignment 11­1. (Mar. 1984).

3. M. R. Branscome, et al. "Evaluation of waste combustion in a dry­process cement kiln at Lone Star Industries, Oglesby, Illinois," Research Triangle Institute draft report on EPA Contract 68­02­3149, Work Assignment 11­1. (Dec. 1984).

4. D. L. Hazelwood, F. J. Smith, E. M. Gartner, "Assessment of waste fuel use in cement kilns," EPA­600/S2­82­013, U.S. Environmental Protection Agency, Cincinnati, Ohio, (Oct. 1982).

5. G. M. Higgins, A. J. Helmstetter, "Evaluation of hazardous waste incineration in a dry process cement kiln," Proceedings of the Eighth Annual Research Symposium: Incineration and Treatment of Hazardous Waste. EPA­600/9­83­003. (Apr. 1983). pp. 243­252.

6. H. C. Jenkins, G. Murchingson, R. C. Adrian, R. D. Fletcher, D. C. Simeroth, "Emissions from Los Robles Kiln," State of California Air Resources Board, Engineering Evaluation Report C­82­080. (Aug. 1983).

7. J. L. Cheney, C. R. Fortune, "Improvements in the methodology for measuring hydrochloric acid in combustion source emissions," J. Environ. Sci. Health A19(3), 337­350 (1984).

8. Federal Register/Vol 54, No. 243, p. 52201 (Dec. 20, 1989).

9. D. Y. MacIver, M. A. Yannone, W. A. Klemm, L. D. Adams, "Reaction products or pseudoparticulate issues in testing portland cement plant emissions," Paper presented at 81st Annual Meeting of Air Pollution Control Association, Texas, (June 19­24, 1988).

Table I

Southdown Cement Kilns Chloride Emission Data

Test.Run Front/ Back Impinger Cl- (ug/sample) NH4+ (ug/sample) NH4+ chloride carrying capacity (ug/sample) Na+ (ug/sample) Na+  chloride carrying capacity (ug/sample) K+ (ug/sample) K+  chloride carrying capacity (ug/sample) Total cation chloride carrying capacity (ug/sample) Measured chloride as% of cation carrying capacity
Florida Plant

1.1F 1560 4720 9290 0 0 9.4 8.52 9300 16.80%
1.1B 31.7 NR 0 NR 0 NR 0 0 ERR
1.2F 1850 2710 5340 0 0 0 0 5340 34.60%
1.2B 37.5 NR 0 NR 0 NR 0 0 ERR
1.3F 378 5400 10600 8.6 13.3 103 93.4 10700 3.50%
1.3B 52.1 81.3 160 14.7 22.7 9.8 8.89 192 27.20%
2.1F 159 1170 2310 36.6 56.4 18.3 16.6 2380 6.70%
2.1B 21.4 NR 0 NR 0 NR 0 0 ERR
2.2F 285 8520 16800 24.2 37.3 8.05 7.3 16800 1.70%
2.2B 154 23 45.3 36.8 56.7 7.35 6.66 109 141.70%
2.3F 2640 17000 33400 791 1220 73.7 66.8 34700 7.60%
2.3B 30.7 54.4 107 0 0 0 0 107 28.70%
Kentucky Plant

1.1F 1280 7650 15100 21 32.4 0 0 15100 8.50%
1.1B 17.5 99 195 49 75.5 0 0 270 6.50%
1.2F 1780 7640 15000 20 30.8 0 0 15100 11.80%
1.2B 16.5 76 150 50 77.1 0 0 227 7.30%
1.3F 2170 8820 17400 17 26.2 0 0 17400 12.50%
1.3B 14.6 106 209 56 86.3 0 0 295 4.90%
Colorado Plant

1.1F 254 115 226 97 150 ND 0 376 67.60%
1.1B 198 31 61.1 329 507 260 236 804 24.60%
1.2F 405 229 451 ND 0 ND 0 451 89.80%
1.2B 21.4 15 29.5 50 77.1 ND 0 107 20.10%
1.3F 448 954 1880 29 44.7 ND 0 1920 23.30%
1.3B ND 34 67 116 179 ND 0 246 0.00%
1.4F 245 1600 3160 37 57 ND 0 3220 7.60%
1.4B 19.5 23 45.3 24 37 ND 0 82.3 23.70%
1.5F 214 410 807 37 57 34 30.8 895 23.90%
1.5B ND 43 84.7 42 64.7 ND 0 149 0.00%
1.6F 272 1230 2420 27 41.6 ND 0 2460 11.10%
1.6B ND 34 67 31 47.8 ND 0 115 0.00%

NR ­ Not Reported

ND ­ Not Detected

Figure 1

Figure 2

Figure 3

Figure 4