GCI TECH NOTES©
12, Number 1
A Gossman Consulting, Inc.
Publication January 2007
There is a growing level of concern
about mercury emissions from cement kilns and interest in the industry
in developing cost effective options for controlling these emissions.
Cement plants have a wide range of mercury inputs and resulting
emissions because of the wide variety of raw materials and fuels used
in the process. Further, the current level of mercury emission control
at cement plants varies from 0% to as high as 95% using existing
particulate control systems. This is the second in a new series
of GCI TechNotes that will examine this issue.
Mercury emissions are regulated based on concern for mercury entering the food chain and bioaccummulating to significant levels that could impact people eating fish. The following is a brief review of the factors that impact the issue of testing modern cement kilns to determine accurate emission rates.
Mercury Emission Testing from Wet Process and Older Straight Dry Cement KilnsOlder wet process and straight dry cement kilns typically remove from the process a small portion of the dust captured in the APCD system on a continuous basis. This process of dust removal combined with essentially steady state operation of the raw feed system allows these older kiln systems to come to fairly quick equilibrium relative to volatile metals - generally within 6-14 hours of any increase or decrease in the metal input rate depending on the size of the surge tank used to store APCD dust prior to feeding back into the kiln. For these reasons it has been possible to test mercury emissions from these kiln systems and produce successful mass balances of the systems for mercury and other trace metals. GCI has a long track record of producing successful trace mass balances on cement kilns. (Papers relating the results of these tests, the mass balance results and guidance on how to perform this sort of mass balance are available in the library at our website.)
When GCI began testing modern precalciner kilns for mercury and other metal emissions in the early 90's we began to see the disturbing trend of poor mass balances on mercury and thallium. Subsequent research into the issue has allowed us to work on the development of a dynamic model that focuses on the feed rate of mercury to the kiln itself during the long term operation of precalciner kilns with in-line raw mills. Because it is typically the case that all of the dust captured in the main APCD is returned to the kiln and because these systems periodically go through a maintenance cycle where the raw mill is shut down and the exhaust gases from the kiln bypass directly to the APCD device, mercury emissions never come to a short term equilibrium and can typically take weeks to come to a long term equilibrium.
Figure 1 illustrates a typical operation Hg input rate where the mercury concentration is approximately 1 ppm relative to the feed and the raw feed rate is about 300 tons per hour. The raw mill maintenance cycle has been modeled as 15 hours out of every 100 hours. Actual operations typically can vary week to week in this regard but for the purposes of understanding the data and the model, randomization of these factors was not deemed necessary. Looking at the model it is easy to see that there is no 3 hour, 12 hour, or even 24 hour period during which the system comes to equilibrium allowing a representative stack test to occur. No wonder the US EPA has expressed confusion over some of the recent stack test results from precalciner cement kilns. It is also worth noting that it takes about 1700 hours (over two months) for the system to reach a long term equilibrium. Only at this point is the long term emissions of mercury equal to the long term input rate. Given the frequency that precalciner kilns are down and require maintenance, the times when the system is in a long term equilibrium will be infrequent.It should also be noted that it is commonly the case that with the raw mill down these systems have higher temperatures in the APCD system and are less likely to have as high a sorption capacity on the dust in the system. This change in capture efficiency of the system would then need to be superimposed on the feed rate, thus resulting in an even more complex and quite likely more variable plot of emission rate.
Attempts to test mercury emission rates from modern precalciner cement kilns are inherently invalid representations based on the non-steady-state operation of these kiln systems relative to mercury input rates and changing capture efficiencies. The establishment by authorities of any short term emission limit and testing requirement can only be considered to be based on "junk science" or a gross misunderstanding of how these systems operate. If limits and testing are to be performed, the limits need to be based on long term input rates combined with capture efficiencies determined by comparing the amount of mercury entering the system with that captured and withdrawn from the thermal processing steps in the cement kiln process.