Five different wastewater COD fractionation methods were employed for simulating an experimental MBR wastewater treatment plant using WEST. The predictions of dynamic simulations using as input the data obtained according to each influent characterization methodology were compared with the results of the experimental system and differences between experimental and predicted values were analyzed in order to select the fractionation method which provides the best fitting and minimizes errors. Three of these methods were based on the determination of the biodegradable fractions using respirometric assays of real wastewater filtered through 0.45- and 0.22-μm pore size filters or adding a previous flocculation step before filtration. Moreover, a method based on physicochemical analyses and another one based on theoretical coefficients were also compared. Simulated system performance and effluent quality greatly depended upon the influent characterization and the proper model calibration. Thus the importance of selecting a suitable fractionation methodology is high, especially in MBR systems working at specific operational conditions that may alter COD fractions. In this study, MLSS in the bioreactors and sludge supernatant COD concentrations were better predicted when the influent characterization was based on respirometric methods. Both the method based on theoretical coefficients and the physicochemical method underestimated the particulate inert fraction and therefore, also the MLSS concentrations. Moreover, these results showed that for a correct effluent COD prediction in MBR systems, it is necessary to take into account that the membrane retained part of the soluble inert fraction.
Keywords: Aerobic processes; COD fractionation; dynamic simulation; membrane bioreactors; modelling; wastewater characterization.