Artificial Neural Network Modeling for Basic Dye Adsorption onto Montmorillonite

Abstract

This research focused on the use of montmorillonite (Mt) clay to eliminate methylene blue (MB) from water and the modeling of the process with an artificial neural network (ANN). The effects of initial MB concentration, contact duration, pH, and temperature on the adsorption efficiency were examined. Maximum adsorption capacity, with 91.3 wt% MB removal at the selected optimum parameters (pH: 12, contact duration: 90 min, temperature: 25 °C, adsorbent dosage: 2 g/L, and initial MB concentration (Ci ): 650 mg/L), was measured as 296.6 mg/g. Langmuir isotherm and pseudo-second-order kinetic models showed the highest correlation coefficients (0.9935 and 0.9999, respectively) to explain the mechanism of the adsorption of MB cations onto Mt. In addition, the maximum adsorption capacity at 25 °C calculated by using the Langmuir isotherm model was determined as 400 mg/g. In the developed ANN model, pH, C i, temperature, and time were considered the input data, and the equilibrium MB concentration (C e) was the output data. While obtaining the ANN model, the data were processed using fivefold cross-validation. Comparing the estimation results with the experimental studies, R 2 = 0.9987 and mean square error (MSE) = 0.00084654 were obtained. Therefore, the C e value estimations were performed with high success rates by using the developed ANN model.