A blend of ferrous chloride and ferric chloride (FeCl 2 -FeCl 3 ) was simultaneously dosed into an activated sludge system at pilot scale in order to test the effect on biological P removal. Additional removal due to chemical precipitation was measured as the difference in system P removal between parallel test and control systems. Both systems strongly exhibited biological excess P removal (BEPR). The extent of P release in the anaerobic reactors of the two systems was compared by mass balance, as one indicator of the relative "magnitude" of BEPR. Phosphorus fractionation of the mixed liquor also served as an indicator of the biological and chemical mechanisms. Evidence was found that the BEPR mechanism is partially inhibited by simultaneous FeCl 2 -FeCl 3 addition, even in the absence of effluent phosphate limitation. However, the degree of inhibition was relatively low, ranging from 3 to 25% (approximately) for Fe doses in the range ca. 10 to 20 mg/l as Fe, with an average system P removal of 14 to 18 mgP/l in the control. FeCl 2 -FeCl 3 dosing in this range was sufficient to produce additional P removal of the order of 1 to 8 mgP/l over periods of one to seven sludge ages per experimental period, depending on the experimental conditions. Sustained operation of the BEPR mechanism in the presence of FeCl 2 -FeCl 3 was possible over a continuous period of seven sludge ages, under conditions in which effluent phosphate was at least partially limiting. Under such conditions, the chemical and biological mechanisms appear to be "disadvantaged" to approximately the same extent, as evidenced by the apparent stoichiometry of Fe:P for the chemical precipitation and magnitude of the poly P containing fractions measured for the biological mechanism. This suggested that the biological mechanism is able to compete effectively with the chemical mechanism under conditions of low reactor phosphate concentrations (~1 mgP/l orthoP) for sustained periods. However, the presence of simultaneous chemical precipitant significantly reduces the extent to which the biological P removal potential is utilised under P-limiting conditions. This could explain the difficulty sometimes reported in the control of full-scale activated sludge systems with simultaneous precipitant addition where a very low effluent P concentration (<1 mgP/l) has to be achieved.

The IAWQ Activated Sludge Model (ASM) No. 2 is a kinetic-based model and incorporates two simple processes for chemical precipitation and redissolution that are readily integrated with biological processes for carbon, nitrogen and phosphorus removal. This model was applied to experimental data collected as part of this study from parallel pilot-scale 3-stage Phoredox systems with and without simultaneous dosing of chemical precipitant. The precipitants tested were alum, ferric chloride and ferrous-ferric chloride. The model was calibrated to the control unit (without precipitant addition) in order to match effluent phosphate (P) predictions (and hence P removal) as closely as possible. The same calibration was then applied to modelling the test unit (with precipitant addition). It was found that the default model input stoichiometry for the precipitation reaction (ideal 1:1 molar ratio of metal ion (Me) to P) was suitable for ferric chloride addition at a 20 d sludge age, but did not accurately reflect the test system behaviour for all experimental periods. A lower stoichiometry (0.60 to 0.75) was required for alum at a 20 d sludge age, and for a blend of predominantly ferrous chloride at a 10 d sludge age. The input stoichiometry was further decreased under P-limiting conditions. A simple approach to, and possible reasons for, the manipulation of the model stoichiometry are discussed in the light of observed stoichiometry from system P removal and fractionation data collected as part of this study. Furthermore, an alternative approach based on manipulation of the precipitation (and hence redissolution) kinetic constant is suggested and evaluated using available experimental data. Model predictions and observed data in respect of polyphosphate (polyP) and suspended solids are also compared and discussed. It is concluded that the ASM No. 2 model provides a useful basis for modelling simultaneous P precipitation, provided certain minor modifications are made. Further investigation into the kinetics of the precipitation process(es) is recommended, particularly in relation to the effect of system sludge age. The model lends itself to further enhancement by incorporating additional physico-chemical processes. Nomenclature

This paper reviews three published models for simultaneous chemical phosphorus precipitation in activated sludge systems using metal salts. In the first, a chemical equilibrium approach is used, based on observations made from batch and continuous-flow tests, a theoretical formula for metal (e.g. ferric) hydroxy-phosphate and a set of metal phosphate complexes or ion pairs for dissolved orthophosphate (orthoP) species. Apart from applying the precipitation stoichiometry observed in admixture with activated sludge, in this model no interaction between the chemical and biological mechanisms is accounted for and no biological processes are modelled. In the second model, a combined equilibrium-kinetic approach is used to model the chemical and biological processes. The chemical and biological processes become kinetically linked through soluble orthoP as a variable. This model includes biological processes for conventional activated sludge systems, but does not include biological excess P removal processes (BEPR). Apart from this limitation, a potential problem in the combined equilibrium-kinetic approach was identified: The precipitation reactions were modelled based on equilibrium chemistry and assumed to be complete at the start of simulation; precipitate, therefore, could not form dynamically during the ensuing kinetic simulation. Furthermore, the model predictions were very sensitive to the choice of certain key equilibrium (or solubility product) constants. The third approach was to model the precipitation (and dissolution) reactions as kinetic processes within a fully kinetic model for activated systems, including the processes for BEPR. This approach depends on the appropriate selection of rate constants for the forward (precipitation) and reverse (dissolution) reactions. In effect, a number of reactions from equilibrium chemistry are combined and replaced with one "surrogate" reaction having its own apparent equilibrium constant. The kinetic approach offers a number of advantages but is still subject to the limitation that it requires calibration against actual data from activated sludge systems in which simultaneous precipitation is applied. Moreover, interaction between the chemical and biological P removal mechanisms in the model is confined to "competition" for available soluble orthoP. This aspect requires further examination.