Vertical flow constructed wetlands using expanded clay and biochar for wastewater remediation: A comparative study and prediction of effluents using machine learning

Abstract

This study evaluated and compared the performance of two vertical flow constructed wetlands (VF) using expanded clay (VF₁) and biochar (VF₂), of which both are low-cost, eco-friendly, and exhibit potentially high adsorption as compared to conventional filter layers. Both VFs achieved relatively high removal for organic matters (i.e. Biological oxygen demand during 5 days, BOD₅) and nitrogen, accounting for 9.5 - 10.5 g^.BOD₅^.m^-2.d^-1 and 3.5 - 3.6 g^.NH₄-N^.m^-2.d^-1, respectively. The different filter materials did not exert any significant discrepancy to effluent quality in terms of suspended solids, organic matters and NO₃-N (P > 0.05), but they did influence NH₄-N effluent as evidenced by the removal rate of that by VF₁ and VF₂ being of 8₂.4 ± 5.7 and 84.6 ± 6.4%, respectively (P < 0.05). The results obtained from the designed systems were further subject to machine learning to clarify the effecting factors and predict the effluents. The optimal algorithms were random forest, generalized linear model, and support vector machine. The values of the coefficient of determination (R²) and the root mean square error (RMSE) of whole fitting data achieved 74.0% and 5.0 mg^.L^-1, 80.0% and 0.3 mg^.L^-1, 90.1% and 2.9 mg^.L^-1, and 48.5% and 0^.5 mg^.L^-1 for BOD₅_VF₁, NH₄^-N_VF₁, BOD₅_VF₂, and NH₄-N_VF₂, respectively.

Keywords: Biochar; Constructed wetland; Expanded clay; Machine learning; Vertical flow.