Performance evaluation and neural network modeling of trichloroethylene removal using a continuously operated two-phase partitioning bioreactor

Abstract

Abstract Biological reactors have widely used in waste gas treatment. The inclusion of non-biodegradable organic solvents (silicone oil) to these reactors is a way to improve/enhance the removal capacity of the biological systems. This study aimed to evaluate the performance of a two-phase continuous stirred tank bioreactor (CSTB) for the removal of trichloroethylene (TCE) by Rhodococcus opacus. The effect of inlet TCE concentration in the range 0.3–3.44 g m−3 on TCE removal studied for 77 days continuously in the CSTB system. During the operation, complete TCE removal has obtained for a high TCE concentration of 1.12–2.62 g m−3 due to the addition of silicone oil as the non-aqueous phase. The maximum elimination capacity was found 372.55 g m−3 h−1 at the inlet TCE loading rate of 423.36 g m−3 h−1. The carbon dioxide (CO2) concentration profile and biomass growth profile from the CSTB revealed that the robustness of the oleaginous bacteria R. opacus in TCE removal. The artificial neural network (ANN) based model was able to successfully predict the performance of the bioreactor using the Levenberg–Marquardt (LM) back propagation algorithm. The prediction accuracy of the model was high as the experimental data and is in good agreement (R 2 = 0 .9923) with the ANN predicted data. This is the first report that used ANN to model a two-phase CSTB treating high TCE polluted stream. Overall, the results indicated that the addition of silicone oil could efficiently improve TCE removal in a two-phase continuous bioreactor.