Determination of the apparent ozonation rate constants of 1:2 metal complex dyestuffs and modeling with a neural network

In this study, the apparent ozonation rate constants of 1:2 metal complex dyestuffs under different empirical conditions such as dye concentrations (400–1000 ppm), ozone–air flowrates (5–15 l min^−1), the percentages of O3 in the ozone–air flowrate (0.7–1.4),pH(3–12), temperatures (18–70^ ◦C), powder activated carbon (PAC) (0.5–1.5 g in solution of 250 ml), HCO3− (0–26 mM) and H2O2 concentrations (0–21 mM) were determined. The ozonation of 1:2 metal complex dyestuffs was found to be fit pseudo-first-order reaction, and the apparent rate constants did not change with the increase of dyestuffs concentrations. For 1:2 metal complex dyestuffs, the apparent rate constants of dyestuffs degradation by ozonation increased with the augmentation of initial pH, H2O2, the percentage of O3 in the ozone–air flow rate and PAC dosage in the solution, but decreased with the increase of HCO3(−) concentration and temperature of the solution. The apparent rate constant of dyestuffs degradation by ozonation increased with the augmentation of ozone–air flow rate from 5 to 10 l min^−1, but it did not change in the range of 10–15 l min^−1. At a high pH, the ozonation of 1:2 metal complex dyestuffs contributed to the increase the apparent rate constant due to the occurrence of hydroxyl free radicals. Using Arrhenius equation, the activation energy (Ea) of the reaction was found as 3 kJ mol−1. The reaction of the ozonation of the dyestuffs under the different temperatures (291, 313 and 343 K) was defined as diffusion controlled according to Ea. The model based on artificial neural network (ANN) could predict the concentrations of the dyestuffs removal from the aqueous solution during ozonation under the different conditions. A relationship between the predicted results of the designed ANN model and the experimental data was also conducted. The ANN model yielded a determination coefficient of R2 = 0.978, a standard deviation ratio of 0.146, a mean absolute error of 19.503 and a root mean square error of 56.600.