Data-based modeling and optimization of a hybrid column-adsorption/depth-filtration process using a combined intelligent approach

Abstract

Lack of robust techniques for optimization of hybrid separation systems is obvious in the literature. A novel hybrid approach for modeling and optimization of a hybrid process consisting of fixed-bed adsorption column (FBAC) and dead-end filtration (DEF) for the removal of methylene blue from water was presented. Artificial neural network (ANN), response surface methodology (RSM) and genetic algorithm (GA) were used for this purpose. ANN was employed to successfully approximate the breakthrough curves. Central composite design was used to investigate the impact of the operating variables, i.e. feed flowrate, initial concentration, adsorption bed length, and filter type on the removal rate as the objective function. RSM variables together with the operation time were considered as input parameters and the normalized concentration was defined as output of the network. A regression model was developed using RSM to correlate the experimental data. The model was then optimized via GA to maximize the removal rate at break time. The optimal removal rate was 3.79 s−1 corresponding to OS-100 filter, 1.12 L/min flowrate, 30 cm bed length and, 40.49 mg/L initial concentration. Sobol sensitivity analysis indicated that flowrate with 43% and filter type with 41% were the most effective variables on the process performance.