Wastewater Treatment Optimization via Experimental Design, Prediction and Analysis

The wastewater treatment process consists of multiple stages, including physical, chemical, and biological treatments, aimed at removing contaminants such as organic matter, nutrients, pathogens, and heavy metals. Integrating artificial intelligence, statistical modeling, and optimization algorithms can improve prediction and control mechanisms, resulting in more sustainable and cost-effective wastewater management solutions. In the literature, two operating factors (pH and coagulant dosage) were simultaneously used to optimize the multi-response (Turbidity, Total Suspended Solids (TSS), Biochemical Oxygen Demand (BOD) and Chemical Oxygen Demand (COD)) problem, with low and high levels (-1,+1) applied during the experimental design phase, and the second-order regression model (Quadratic Model (QM)) used for the fitting. However, the existing work was insufficient in accurately fitting the data due to the selection oaf factors, lack of variable bandwidths and the chosen regression model. In this paper, we introduce an additional critical factor to the existing set of factors and extend the levels to five, aiming to achieve a better fit for data and optimally maximize the response in line with process requirements. The model applied is aimed to enhance goodness-of-fit statistics, achieve dispersion from the zero residual line, and improve optimization results. Comparative analysis, including 3D surface plots, demonstrated that the local linear regression (〖LLR〗_AB) model with variable bandwidths outperformed the existing QM in terms of design technique. The 〖LLR〗_AB model exhibited superior goodness-of-fit statistics, minimized residual errors in terms of predicted response, and achieved better optimization results, with an overall desirability of 99.5% compared to 98.6% for QM. Keywords: Local Linear Regression Model, Citrullus Lanatus, Turbidity, Total Suspended Solids, Biochemical Oxygen Demand, Chemical Oxygen Demand. Aims Research Journal Reference Format: Odigie O. O. & Eguasa O. (2024): Wastewater Treatment Optimization via Experimental Design, Prediction and Analysis. Advances in Multidisciplinary Research Journal. Vol. 11 No. 1, Pp 17-33. www.isteams.net/aimsjournal. dx.doi.org/10.22624/AIMS/V11N1P2