Comparative Performance Analysis of Aluminium-based and Iron-based Coagulants: Modelling Using Artificial Neural Networks

¹ Department of Civil Engineering, TSSM’s Bhivarabai Sawant College of Engineering and Research, Pune 411041, India
² Department of Civil Engineering, JSPM’s Narhe Technical Campus, Pune 411041, India
³ Saudi Investment Recycling Company, Riyadh 14223, Saudi Arabia

^* Corresponding author: This email address is being protected from spambots. You need JavaScript enabled to view it.

Abstract

Greywater reuse is increasingly promoted to alleviate freshwater stress; however, effective and statistically validated treatment strategies are required to reduce contamination. This study offers the first integrated experimental statistical artificial neural network (ANN)-based comparison report between major aluminium-based (alum and polyaluminium chloride, PAC) and iron-based (ferric chloride and ferrous sulphate) coagulants. These are used for greywater treatment with the support of analysis of variance (ANOVA) with Dunn-Sidak's significance (α = 0.05) test and modelling with ANN. All reported removal efficiencies in the present study are calculated from mean removal efficiencies with associated standard deviations. Although treatment performance for aluminium and iron-based coagulants showed no significant difference in terms of overall performance with respect to turbidity removal (p > 0.05), polyaluminium chloride (PAC) showed the highest turbidity and TSS removals (94.8% and 90.4% respectively) at a lowest average dose (122 ± 26 mg/L) whereas alum and iron-based coagulants, at comparable dosages, required a higher dosage to achieve a similar amount of particulate removal. PAC showed BOD, COD and ammoniacal-nitrogen removals of 71.2, 76.8 and 52.7%, respectively. Phosphate and faecal coliform removals were more than 99.0% and 1.5 log with all coagulants, respectively. The ANN models are able to predict optimum coagulant dose and treated greywater quality with correlation coefficients (R) of 0.88-0.91 for dosage prediction and 0.99 for treated greywater quality, respectively, which indicates the reliability and robustness of the modelling approach. These correlation coefficients have been produced from entirely separate testing and validation test data, which validates robust model generalisation. All in all, PAC is indicated for decentralised greywater reuse systems, because of its lower chemical dosing and stable treatment characteristics, which result in a lower chemical consumption, consistent treatment behaviour and a simplified handling requirement.