Artificial neural network (ANN) approach for prediction and modeling of breakthrough curve analysis of fixed-bed adsorption of iron ions from aqueous solution by activated carbon from Limonia acidissima shell

Abstract

Abstract The present research article explored the potential of activated carbon prepared from Limonia acidissima shell to adsorb total Fe ions from aqueous solution in a packed bed up-flow column. The effect of essential factors such as bed height (3–5 cm), initial concentration (30–50 mg/L), and flow rate (3.32–5.4 mL/min) on the performance of the column bed was investigated. The adsorption capacity augmented with an increase in bed height and initial adsorbate concentration but declined with an increase in flow rate. The maximum uptake capacity of 209.6 mg/g was achieved at 5 cm bed height, 3.32 mL/min, and 50 mg/L initial concentration. The bed depth service time (BDST) model was used to analyze the experimental data and determine the characteristic parameters of the packed bed reactor suitable for designing large-scale column studies. The Adams–Bohart, Thomas, and Yoon–Nelson models were applied to the experimental data to predict breakthrough curves using non-linear regression. The artificial neural network (ANN) based model was able to efficaciously predict the column performance using the Levenberg–Marquardt (LM) algorithm. A comparison between the experimental data and model results contributed to a high degree of correlation, specifying that the preliminary information was in good agreement with the ANN predicted data.