Artificial Neural Network Modelling of Photocatalytic Degradation of Diclofenac as a Pharmaceutical Contaminant

Abstract

In this work, the photocatalytic removal of diclofenac (DCF) was investigated using TiO2-P25 nanoparticles immobilized on glass beads in a packed bed photoreactor. DCF is one of the non-steroidal anti-inflammatory drugs used as an analgesic drug. DCF is monitored in urban sewage and surface waters as a stable contaminant that can harm the environment. Advanced oxidation processes (AOPs) are promising methods for degradation and removal of environmental pollutants. The heterogeneous photocatalysis process is one of the AOPs so that the contaminants are decompose in the presence of UV light and a photocatalyst (TiO2). Holes and hydroxyl radicals are the main active species in the UV/TiO2 process. A thin layer of TiO2-P25 nanoparticles was immobilized by heat attachment method on glass beads. The effect of five operational parameters, the initial concentration of DCF, the power of the light source, the flow rate of the fluid in the photoreactor, irradiation time and pH, has been studied experimentally in the efficiency of the photoreactor. The DCF removal percent is 99% for the initial DCF concentration of 10 mg L–1, the power of the light source of 16 W, the fluid flow rate of 240 mL min–1 and pH 6 for 120 min irradiation time. The effect of operational parameters on the DCF removal percent was modeled using the artificial neural network (ANN). ANN modeling with a 5 : 9 : 1 feed-forward back propagation neural network demonstrated the appropriate consistency of the experimental and predicted data. The R2 values for all data (training, validation and test) were close to 1, confirming ANN reasonable predictive performance. Using the weights of the ANN model in the Garson equation, indicated that pH and irradiation time had the highest effect on the DCF removal percent.