Mass balance of toxic metals in cement and aggregate kilns co-fired with fossil and hazardous waste-derived fuels

The co-firing of conventional fossil fuel with hazardous waste-derived fuel (WDF) in cement and aggregate kilns has increased considerably since 1984. Data are compiled from compliance-test reports for cement and light-aggreg ate kilns at steady-state conditions. These data reveal that the majority o f each metal is incorporated into the kiln dust and product (cement clinker or aggregate product). Distribution ratios, for kiln dust and emissions re lative to the total kiln system, are calculated for the metals arsenic (As) , beryllium (Be), cadmium (Cd), chromium (Cr), and lead (Pb). Calculations, which use these ratios, balance the input and output metal mass by assigni ng the remaining metal to the product. These balance calculations include k ilns that recirculate kiln dust and those that do not. Comparing reported a nd calculated metal concentrations in the product (cement clinker or aggreg ate product) provides a qualitative method for evaluating metals balance. M ost compliance data yield poor agreement between the input and output masse s. Metal distributions in kilns that recirculate different portions of ceme nt kiln dust (CKD) indicate an increased tendency to concentrate As and Cr into CKD with increased CKD recirculation. This effect likely results from the rather low volatility of As and Cr. Metal concentrations in CKD, which are computed for the co-firing of WDF and fossil fuel in the kilns, are dis tinctly higher than those based on burning fossil fuel alone. A moderate to strong correlation of CKD metal concentrations with fuel metal concentrati ons indicates a fundamental control of CKD composition by fuel composition. Metal concentrations calculated for the transient approach to steady-state conditions provide a theoretical representation of that process. Equilibra tion test data, from compliance reports, show an irregular approach to stea dy-state conditions. This variable accumulation and release behavior appear s typical of the transient part of the process, and may increase the uncert ainty in assessing steady-state conditions.