Abstract: Increases in the treatment of water to meet the growing water demand ultimately result in unmanageable quantities of residuals, the handling, and disposal of which is a major environmental issue. Consequently, research into beneficial reuse of water treatment residuals continues unabated. This study investigated the applicability of lime-iron sludge for phosphate adsorption by fixed-bed column adsorption. Laboratory-scale experiments were conducted at varying flow rates and bed depths. Fundamental and empirical models (Thomas, Yan, Bohart-Adams, Yoon-Nelson, and Wolboroska) as well as artificial intelligence techniques (Artificial neural network (ANN) and Adaptive neuro-fuzzy inference system (ANFIS)) were used to simulate experimental breakthrough curves and predict column dynamics. Increase in flow rate resulted in reduced adsorption capacity. However, adsorption capacity was not affected by bed depth. ANN was superior in predicting breakthrough curves and predicted breakthrough times with high accuracy (R² > 0.9962). NaOH (0.5 mol·L^-1) was successfully used to regenerate the adsorption bed. After nine cyclic adsorption/desorption runs, only a marginal decrease in adsorption and desorption efficiencies of 10% and 8% respectively was observed. The same regenerate NaOH solution was reused for all desorption cycles. After nine cycles the eluent desorbed a total of 1550 mg phosphate exhibiting potential for further reuse.

Key words: Adsorption, Phosphate, Sludge, Adaptive Neuro-fuzzy Inference System, Neural Network

摘要： Increases in the treatment of water to meet the growing water demand ultimately result in unmanageable quantities of residuals, the handling, and disposal of which is a major environmental issue. Consequently, research into beneficial reuse of water treatment residuals continues unabated. This study investigated the applicability of lime-iron sludge for phosphate adsorption by fixed-bed column adsorption. Laboratory-scale experiments were conducted at varying flow rates and bed depths. Fundamental and empirical models (Thomas, Yan, Bohart-Adams, Yoon-Nelson, and Wolboroska) as well as artificial intelligence techniques (Artificial neural network (ANN) and Adaptive neuro-fuzzy inference system (ANFIS)) were used to simulate experimental breakthrough curves and predict column dynamics. Increase in flow rate resulted in reduced adsorption capacity. However, adsorption capacity was not affected by bed depth. ANN was superior in predicting breakthrough curves and predicted breakthrough times with high accuracy (R² > 0.9962). NaOH (0.5 mol·L^-1) was successfully used to regenerate the adsorption bed. After nine cyclic adsorption/desorption runs, only a marginal decrease in adsorption and desorption efficiencies of 10% and 8% respectively was observed. The same regenerate NaOH solution was reused for all desorption cycles. After nine cycles the eluent desorbed a total of 1550 mg phosphate exhibiting potential for further reuse.

关键词: Adsorption, Phosphate, Sludge, Adaptive Neuro-fuzzy Inference System, Neural Network