Prediction of iron(II) breakthrough in adsorptive filters under anoxic conditions

Physical, chemical and biological mechanisms may contribute to the removal of iron in filters. This paper focuses on the physical–chemical mechanisms namely adsorptive filtration and floc filtration. Iron removal filters operating in physical–chemical mode are expected to perform better in terms of filtrate quality, run time and costs by changing the governing mode of operation from floc filtration to adsorptive filtration. In adsorptive filtration, iron(II) ions adsorb onto iron oxide coating on the filter media and are subsequently oxidised to iron(III). Modelling of adsorptive iron removal is an essential tool to predict the performance of filters operated under adsorptive mode and to optimise design parameters for adsorptive iron removal filters. The iron concentration in filtrates of adsorptive filters with new (fresh, uncoated) and iron oxide coated sand operating under anoxic conditions, was modelled using adsorption isotherm, mass balance and mass transfer equations. Kinetic rate constants, isotherm parameters and mass transfer coefficients used for model predictions were determined from batch and short column experiments. The measured filtrate iron concentrations of laboratory filter columns were then compared with the predictions of three different fixed bed adsorption models. Consideration of external mass transfer alone (constant pattern model, CPM) was not sufficient to predict the increase of filtrate iron concentration with time. Further improvement in prediction was observed when the linear driving force model (LDFM) was used and the effect of dispersion was also included in the model. The inclusion of surface diffusion (plug flow homogeneous surface diffusion model, PFHSDM) improved model predictions significantly compared with the CPM. The LDFM and the PFHSDM predictions of iron breakthrough were better in the case of new sand compared with iron oxide coated sand. The difference in model predictions and experimental results in the case of coated sand was probably due to initial pH drop in the pores of the filter media with iron(II) adsorption, and a consequent decrease in iron(II) adsorption capacity.