Biofilm characterization in removal of total chemical oxygen demand and nitrate from wastewater using draft tube spouted bed reactor.

The present paper investigates the effect of dilution rate on the removal of total chemical oxygen demand and nitrate in the draft tube spouted bed reactor and morphological characteristics of biofilms formed by microorganisms of mixed culture on granular activated carbon (GAC). The nitrate and total chemical oxygen demand (COD) decreased from 97 to 81% and 95% to 87% respectively with increase in dilution rate from 0.6/h to 1.5/h showing that residence time in the reactor governs the nitrate and total COD reduction efficiency. Lower dilution rates favor higher production of biomass and extracellular polymeric substances (EPS). It was observed that the nitrate and total COD reduction rate increased with time along with simultaneous increase in EPS production. Thus, the performance of a reactor in terms of dynamic and steady-state biofilm characteristics associated with nitrate and organic reduction is a strong function of dilution rate. Hence these findings indicate that a draft tube spouted bed reactor is capable of simultaneously reducing total organics and nitrogen in industrial/municipal wastewater, as this reactor possesses two distinct regions aerobic and anoxic conditions which can prevail in different parts of a reactor.

Comparison of response surface methodology and artificial neural network to enhance the release of reducing sugars from non-edible seed cake by autoclave assisted HCl hydrolysis.

In the current investigation, statistical approaches were adopted to hydrolyse non-edible seed cake (NESC) of Pongamia and optimize the hydrolysis process by response surface methodology (RSM). Through the RSM approach, the optimized conditions were found to be 1.17%v/v of HCl concentration at 54.12 min for hydrolysis. Under optimized conditions, the release of reducing sugars was found to be 53.03 g/L. The RSM data were used to train the artificial neural network (ANN) and the predictive ability of both models was compared by calculating various statistical parameters. A three-layered ANN model consisting of 2:12:1 topology was developed; the response of the ANN model indicates that it is precise when compared with the RSM model. The fit of the models was expressed with the regression coefficient R2, which was found to be 0.975 and 0.888, respectively, for the ANN and RSM models. This further demonstrated that the performance of ANN was better than that of RSM.