Artificial neural network and techno-economic estimation with algae-based tertiary wastewater treatment

Abstract

In this study, secondary-treated (ST) wastewater effluents supplemented with different nutrient sources were employed for microalgae cultivation in outdoor pilot-scale pools operated under natural environmental conditions. The BG11-supplemented wastewater showed a high algal biomass concentration = 0.79 ± 0.04 g/L, with NO3−, NH4+, and PO43− removal efficiencies = 83.20 ± 2.90 %, ≈ 100 %, and 93.50 ± 3.28 %, respectively. The corresponding lipid, protein, and carbohydrate contents of microalgae were 25.60 ± 0.80 %, 29.00 ± 0.88 %, and 18.40 ± 1.00 % w/w dry cell weight (DCW) basis, respectively. The ranges of protein and carbohydrate contents of lipid-extracted biomass were 25–40 % and 17–23 %, respectively. A three-layer feed-forward back-propagation artificial neural network (ANN) was used to predict the microalgae DCW, based on six inputs, i.e., temperature, pH, dissolved oxygen (DO), electrical conductivity (EC), NO3−, and PO43-. The optimized ANN architecture (6–10 – 1), with the Levenberg-Marquardt training algorithm, achieved the highest predictive performance (R2: 0.983). Based on the ANN sensitivity analysis, the environmental factor’s relative importance arranged as NO3− > PO43- > pH ≈ DO > temperature > EC. The nutrient removal ability and biochemical composition of microalgae were expressed regarding capital and operational costs, and profits. The payback period of the wastewater-based algal cultivation system was shorter than the project’s lifetime, implying a sustainable and feasible application.