A Bayesian Approach to Experimental Design for Tetracycline Adsorption in a Fixed-Bed Column: A Comparison between Activated Carbon and Modified Activated Carbon

A Bayesian Approach to Experimental Design for Tetracycline Adsorption in a Fixed-Bed Column: A Comparison between Activated Carbon and Modified Activated Carbon

• Autor
• Letícia Reggiane de Carvalho Costa
• Co-autores
• Júlia Toffoli de Oliveira , Gustavo Dall Agnol , Liliana Amaral Féris
• Resumo

•  

  In this study, the efficiency and applicability of fixed-bed columns were evaluated for tetracycline (TC) adsorption on granular activated carbon (GAC) and modified granular activated carbon (GAC-Cu). An experimental design using the Box-Benking (BOX) method was employed to optimize column operation, considering three independent variables: initial pollutant concentration (20, 50, and 80 mg/L), bed height (1, 2, and 3 cm), and feed flow rate (3, 6, and 10 mL/min). Response factors included saturation time (t_sat) and maximum adsorption capacity (q_max). To simulate breakthrough curves, six models (Thomas, Yoon-Nelson, Yan, Clark, Gompertz, and Log-Gompertz) were applied under optimal conditions. Parameter estimation and model selection were performed using Bayesian techniques, incorporating experimental qmax values and five statistical metrics (R² and R²_ajstd; AIC, AIC_C, and BIC). The results demonstrate that both GAC and GAC-Cu exhibit high efficiency in TC removal, achieving 95% removal under optimal operational conditions (C_0TC = 80 mg/L, Q = 6 mL/min, and Z = 1 cm). For CAG, these conditions resulted in a saturation time (t_sat) of 23 minutes and a maximum adsorption capacity (q_max) of 2.08 mg/g, while for GAC-Cu, a significantly higher adsorption bed capacity of 6.87 mg/g was observed in 107 minutes. These findings suggest that the modification of GAC enhances its adsorption capacity for TC, making it three times more effective than unmodified GAC. Additionally, the saturation time indicates the efficiency of the process in treating high TC concentrations with minimal solid mass, highlighting its feasibility for implementation on a larger scale. Applying the Bayesian technique, it was observed that, for GAC and CAG-Cu respectively, the Yan and Clark models were the best predictors of the breakthrough curve behavior, elucidating the mechanisms of the column. Moreover, the Log-Gompertz model was useful for the scale-up of the process. Finally, this work contributes valuable information for the development and implementation of environmentally sustainable and effective water treatment strategies.

   

• Palavras-chave
• tetracycline; fixed-bed column; mathematical modeling; parameter estimation; Bayesian technique
• Área Temática
• Adsorção aplicada ao meio ambiente