Abstract
The removal of hydrochlorothiazide (HCT) from molecular liquids on multi-walled carbon nanotubes (MWCNTs) and single-walled carbon nanotubes (SWCNTs) was studied in a batch system. In this present work, multiple linear regression (MLR) and radial basis function neural network (RBFNN) were utilized to forecast the adsorption removal percentage of HCT by both adsorbents. The influences of process variables on the removal efficiency were optimized by culture algorithm (CA) optimization. The results displayed the RBFNN was better than the MLR to simulate removal of HCT by two adsorbents. The optimal RBFNN model using the test dataset predicted HCT removal (%) with the coefficient of determination (R2) values of 0.8460 and 0.9438; mean squared error (MSE) values of 0.0117 and 0.0010, respectively for SWCNTs and MWCNTs. At the optimum value of parameters, the adsorption isotherms could be fitted well by the Langmuir model with adsorption capacity values of 66.225 mg g−1 for MWCNTs and 45.662 mg g−1 for SWCNTs. It was also found that the pseudo second-order and intraparticle diffusion models were more suitable for explaining the adsorption mechanism by CNTs.
[Display omitted]
•We used SWNT and MWNT for removal of Hydrochlorothiazide.•We model adsorption of Hydrochlorothiazide using radial basis function neural network.•The culture algorithm was used for optimization for Hydrochlorothiazide removal.