Statistical computation and artificial neural algorithm modeling for the treatment of dye wastewater using mucuna sloanei as coagulant and study of the generated sludge

Abstract

Adaptive neuro-fuzzy inference system and response surface methodology techniques were used to predict coagulation-flocculation parameters needed to remove color and COD effectively and produce acceptable sludge using Mucuna sloanei at 298 K from a dye wastewater. The variables input to the network were coagulant dosage (1000, 1400 and 1800 mg/L), solution pH (2, 6 and 10) and stirring time (5, 15 and 25 min). Coefficient of determination, R2 and root mean square, RMSE were used to evaluate the adequacy and predictive relevance of the two techniques. Color removal model indicators are R2 0.9823, RMSE 0.2599 and R2 0.8616, RMSE 21.403 for ANFIS and RSM, respectively; COD removal indicators are respectively, R2 0.9752, RMSE 0.2009, and R2 0.9741, RMSE 0.8118 for ANFIS and RSM; while sludge volume index model indicators are R2 0.9950, RMSE 0.2341, and R2 0.9930, RSME 2.1436, for ANFIS and RSM, respectively. With a limited set of data, the generated models produced idealized findings and were proven to be useful for forecasting color and COD elimination, and SVI. Nevertheless, the ANFIS model is clearly favored because of greater R2 values and lower RMSE.

Keywords

• RSM- ANFIS
• Optimization and modelling
• Coagulation/flocculation
• Mucuna sloanei
• Sludge characterization
• Dye wastewater