Environ. Sci. Technol. 2008, 42, 3717–3723
Assessing Contaminant Mobilization from Waste Materials: Application of Bayesian Parameter Estimation to Batch Extraction Tests at Varying Liquid-to-Solid Ratios S A S C H A C . I D E N , * ,† M A R K U S D E L A Y , ‡ FRITZ H. FRIMMEL,‡ AND WOLFGANG DURNER† Institut für Geoökologie, Technische Universität Carolo-Wilhelmina zu Braunschweig, Germany, and Engler-Bunte-Institut, Chair of Water Chemistry, Universität Karlsruhe (TH), Germany
Received September 11, 2007. Revised manuscript received February 12, 2008. Accepted February 15, 2008.
We investigated the release of chloride, sulfate, sodium, copper, chromium, and dissolved organic carbon from a demolition waste material and a municipal waste incineration product. Batch leaching tests at the liquid-to-solid ratios (L/S ratios) 1, 2, 5, 10, and 50 L kg-1 were carried out and the parameters of a mass balance-partitioning model were estimated from measured concentrations in the extracts by applying a Bayesian approach using a Markov Chain Monte Carlo sampler. We assessed the uncertainty of the model parameters, the desorption isotherms, and the model-predicted concentrations, respectively. Both the excellent fit to the experimental data and a comparison between the modelpredicted and independently measured concentrations at the L/S ratios of 0.25 and 0.5 L kg-1 showed the applicability of the model for almost all studied substances and both materials. Since experimental difficulties impede extraction tests at L/S ratios representative of field soil–water contents, the predictability of concentrations in this range is of great practical relevance for risk assessments. We conclude that batch extraction tests at varying L/S ratios provide, at moderate experimental cost, a powerful complement to established test designs like column leaching or single batch extraction tests.
Introduction The high production of waste materials is a worldwide problem (1, 2). Apart from avoiding waste production, one of the main aims of waste management policy is recycling and reuse of waste materials, e.g., as fill materials applied to natural ground (3). In order to assess the related environmental risk, it is necessary to investigate the release of inorganic and organic compounds from these materials. For this purpose, numerous types of laboratory leaching tests have been developed and discussed (4–8). Basically, the laboratory tests can be categorized into batch leaching and column leaching tests (9, 10). Batch leaching tests use * Corresponding author phone: +495313915602; fax: +495313915637; e-mail:
[email protected]. † Technische Universität Carolo-Wilhelmina zu Braunschweig. ‡ Universität Karlsruhe (TH). 10.1021/es702171z CCC: $40.75
Published on Web 04/18/2008
2008 American Chemical Society
both different extractants (eluents) and liquid-to-solid ratios (L/S ratios). This is partly due to a lack of harmonizing guidelines at the international scale (5, 11), but also because of the fact that different tests aim to give information on different aspects of contaminant release from waste materials. In general, at least three key variables are of interest (12): (i) the liquid-phase concentration of a contaminant occurring in situ, (ii) the total amount of a contaminant that will be leached from the waste material in the long term, and (iii) the temporal development of the emitted concentration. The latter aspect is often approached by column leaching tests yielding functional relationships between the concentration in the liquid phase and the cumulative L/S ratio (9). As an alternative, it has been shown that batch leaching tests at different L/S ratios provide sufficient information to statistically estimate contaminant release parameters (12). Formally, one can distinguish two contrasting types of relationships between the aqueous phase concentration and the L/S ratio: If solubility controls the concentration in the aqueous phase or the compound under study undergoes very strong sorption, the liquid phase concentration will be constant with changing L/S ratio (12, 13). If sorption is less pronounced, the desorption process will reduce the concentration in the solid phase and the liquid phase concentration will decrease with increasing L/S ratio (12). A quantitative description of the relationship between the aqueous phase concentration and the L/S ratio can be obtained by fitting empirical functions to the measured data. The resulting function parameters can be linked to material properties by statistical approaches similar to the pedotransfer function approach in soil science (14). However, due to its empirical nature, this technique impedes an interpretation of the results in terms of physical or chemical material characteristics. A mechanistic interpretation of contaminant release as function of the L/S ratio requires a detailed geochemical simulation. Such an approach hinges on information about the mineral composition of the solid phase, which increases the costs of the investigation. Moreover, thermodynamic data needed for solving the resulting systems of equations are hardly available for complex cement phases that are abundant in waste materials (15). A reasonable and practical compromise is to simulate the partitioning of the contaminant between the liquid and the solid phase by a linear or nonlinear isotherm and estimate parameters of the distribution model to describe the data (12). The objective of this article is to assess the release of selected compounds (chloride, sulfate, chromium, copper, sodium, and dissolved organic carbon) from a demolition waste material (DWM) and a municipal waste incineration product (MWIP) (9, 16) and to simulate the relationship between the liquid-phase concentration and the L/S ratio by a simple partitioning model. Bayesian parameter estimation is used to identify isotherm parameters and the waterextractable contents of the compounds from observed liquidphase concentrations of batch leaching tests at varying L/S ratios. Particular focus is devoted to predict concentrations at L/S ratios representative for in situ conditions. This is of great practical relevance because liquid extraction at typical field soil–water contents is experimentally hard to achieve. In practice, extraction tests at small L/S ratios (
