Artificial Neural Networks, Optimization and Kinetic Modeling of Amoxicillin Degradation in Photo-Fenton Process Using Aluminum Pillared Montmorillonite-Supported Ferrioxalate Catalyst

Abstract

An artificial neural network (ANN) was applied to study the hierarchy of significance of process variables affecting the degradation of amoxicillin (AMX) in a heterogeneous photo-Fenton process. Catalyst and H2O2 dosages were found to be the most significant variables followed by degradation time and concentration of AMX. The significant variables were optimized and the optimum condition to achieve degradation of 97.87% of 40 ppm AMX was 21.54% excess H2O2 dosage, 2.24 g of catalyst in 10 min. A mathematical model (MM) for the degradation of AMX was developed on the basis of the combined results of the ANN and the optimization studies. The MM result showed that increases in both catalyst and H2O2 dosage enhanced the rate of AMX degradation as shown by the rate constants evaluated from the model. The highest rate constant at the optimum conditions was 122 M–1 S–1. These results provided invaluable insights into the catalytic degradation of AMX in photo-Fenton process.

Optimization of catalytic hydrodeoxygenation of oleic acid into biofuel using fluoroplatinum oxalate zeolite supported catalyst

“…9.24 for model Y and 8.84 for model Z). The Model F-value is 77.77 and 15.30 for the response function Y and Z, respectively, and these values implied that the models are significant and there is less than 0.01% probability that Model F-values this large could occur due to model error or noise, although it may be associated to experimental error [40][41][42].…”

“…Similarly their adjusted regression coefficient, R 2 adj , that corrects regression coefficient, R 2 , based on the number of terms in the models and the sample size [41] are 0.9782 and 0.9548, respectively. Generally, if a model contains many terms and has small sample size this could result in a much lower R 2 adj compared to R 2 [40,41], in these cases the models have high R 2 and R 2 adj which indicated that the chosen quadratic model for response surface method (RSM) adequately describe the experimental data in the range of the operating parameters. Furthermore, the ''Pred R-Squared'' of 0.8511 and 0.8325 are in reasonable agreement with the ''Adj RSquared'' of 0.9782 and 0.9548 for the HDO and ISO models, respectively.…”

“…7c and d) are the most reliable source of information. For example, the plot of run numbers vs. outliers T points had been used to evaluate the assumption of constant variance [41]. It can be seen from Fig.…”

Application of Artificial Neural Networks for Catalysis: A Review

“…Basri et al also found that, for optimizing the lipase-catalyzed synthesis of a palm-based wax ester, ANN superiorly outperformed RSM [58]. More comparative studies using RSM and ANN can be found in References [59][60][61][62][63][64][65].…”

Treatment of amoxicillin by O 3 /Fenton process in a rotating packed bed

“…Fenton (H 2 O 2 and Fe 2þ ) and the related processes are easy-to-handle AOPs that have been used for antibiotic wastewater treatment (Ayodele et al, 2012;Elmolla and Chaudhuri, 2009a;Elmolla et al, 2010;Trov o et al, 2011). For instance, Elmolla and Chaudhuri (2009b) treated amoxicillin solution with Fenton reagent and a complete degradation of amoxicillin in 2 min and 81.4% of COD removal in 10 min were attained.…”