* Corresponding author. 1944-3994/1944-3986 © 2017 Desalination Publications. All rights reserved. Desalination and Water Treatment www.deswater.com doi: 10.5004/dwt.2017.21438 92 (2017) 72–79 October Kinetic study and modelling of cephalexin removal from aqueous solution by advanced oxidation processes through artifcial neural networks Otidene R.S. da Rocha a, *, Renato F. Dantas b , Welenilton José do Nascimento Júnior a , Yuji Fujiwara a , Marta Maria Menezes Bezerra Duarte a , Josivan Pedro da Silva a a Departament of Chemical Engineering, Federal University of Pernambuco, Artur de Sá s/n, Cidade Universitária, Recife, Brazil, Tel. +55 81 21267295; emails: otidene@hotmail.com (O.R.S. da Rocha), welenilton@gmail.com (W.J. do Nascimento Júnior), yujif85@gmail.com (Y. Fujiwara), mmmbduarte@gmail.com (M.M.M. Bezerra Duarte), josivan_silva@hotmail.com (J.P. da Silva) b School of Technology, University of Campinas – UNICAMP, Paschoal Marmo 1888, 13484-332, Limeira, Brazil, email: renatofalcaod@ft.unicamp.br (R.F. Dantas) Received 29 March 2017; Accepted 10 September 2017 abstract The degradation of the antibiotic cephalexin (CEX) was studied by UV direct photolysis and hydrogen peroxide combined with UVC and solar radiation. A factorial plan was used to evaluate the efficiency of the processes and the influence of variables. UVC direct photolysis had a minor contribution (12%) on CEX removal during the UV/H 2 O 2 treatment. The best UV/H 2 O 2 performance from the factorial plan was able to achieve a high degradation percentage for CEX and aromaticity (83.2% and 76.2%, respectively) in 60 min, while solar photolysis was not able to achieve high degradation percent- age at the applied conditions. Statistical analyses pointed to the high statistical significance of the oxidant concentration for the process and the weak dependence of the other variables. The kinetic study demonstrated that the pseudo-first-order model was the more appropriate for both direct pho- tolysis and UV/H 2 O 2 treatments with rate constants of k UVC = 0.0031 min −1 and k UV/H 2 O 2 = 0.0367 min −1 . The use of artificial neural network was proven to be efficient to predict CEX removal by photolysis and photochemical treatments from aqueous solutions. Keywords: Advanced oxidation processes; Cephalexin; Photochemical oxidation; Photolysis; Artificial neural networks 1. Introduction The contamination of water bodies by different types of organic pollutants is an increasing concern among researchers around the world [1]. Due to the current legislations, the demand for high-quality water and effluents has increased [2]. However, some pollutants such as pharmaceuticals and personal care products have shown a high potential of persistence in different aquatic environments, being recognized in the last few years as Contaminants of Emerging Concern because of their acute toxicity and other potentially harmful effects [3–5]. Thus, many studies were focused on their toxicology [6] while others were devoted to develop alternative removal methods [7,8]. Large amounts of antibiotics have been produced in the last century [9], and their use has led to a major problem: bacterial resistance. Antibiotic residues generated by incom- plete metabolism of humans and animals are released into the environment as excretion, causing and enhancing bacte- ria resistance [10–12]. The conventional water treatment is capable of removing only about 5% of their accumulated res- idues [13] since they are highly resistant to the conventional biological degradation methods [14,15]. Cephalosporins are a group of antibiotics in use since 1964, which present different antimicrobial activities. Their efficiency is associated with their low toxicity, simple utiliza- tion and effective pharmacokinetic properties [16,17]. Among this group, a semisynthetic antibiotic called cephalexin (CEX) is the largest applied cephalosporin in the world [14].