Predicting Residual Adsorbable Organic Halides Concentrations in Industrial Wastewater Using Typical Wastewater Parameters

摘要

The aim of this study was to predict the residual adsorbable organic halides (AOX) concentration in an industrial wastewater using conventional, easy-to-measure wastewater parameters. In a pilot test unit, the wastewater was subjected to ozonation at various intensities, resulting in an AOX-removal and hence varying AOX concentrations. In first instance, the parameters used for modeling were selected using Pearson and Spearman correlations. Secondly, multiple linear regression (MLR) was used as a modeling tool to predict both the soluble and total AOX concentration in wastewater samples. To prevent overfitting, a 10-fold cross-validation was carried out. It was found that both the soluble and the total AOX concentration can be predicted using typical wastewater parameters. The measured parameters were pH, chloride concentration, Water-Soluble Organic Carbon concentration (WSOC), UV-VIS spectrum, turbidity, and Solids Removable by Filtration (SRF). Out of these parameters, the following parameters were found to be significant for prediction of the total AOX concentration: turbidity; SRF; UV-VIS absorbance at 200; 227, and 250 nm; and pH. UV-VIS absorbance at 200 and 227 nm and turbidity of the wastewater were found to contribute significantly to the final model. For the soluble AOX concentration, the significant parameters were turbidity; SRF; absorbance at 200, 227, and 250 nm; pH, and chloride concentration. Here, UV-VIS absorbance at 200 and 227 nm were found to contribute significantly to the final model. The obtained final models had an adjusted R2 of 0.921 and 0.916 for the total and soluble AOX, respectively. As a result of the obtained models, both AOX concentrations can be predicted using parameters that are easier to determine. This allows for a significant reduction in wastewater sampling and analysis time and offers the opportunity to optimize the ozone dosing in the wastewater treatment process in the future.