Response Surface Methodology and Artificial Neural Network Modeling for the Removal of Volatile Organic Compounds in Biotrickling Filters

Volatile organic compounds (VOCs) emission from electronic manufacturing industries poses severe air pollution and damage to human health; a cost-effective control strategy is urgently needed. Herein, biotrickling filters (BTF) packed with the polyhedron hollow polypropylene balls were employed to remove VOCs emitted from an electronic manufacturing industry. Response surface methodology with Box-Behnken design (RSM-BBD) and backpropagation artificial neural network coupled genetic algorithm model (GA-BPANN) was proposed to simulate and predict BTF performance on VOCs removal at different inlet VOCs concentrations, empty bed retention times (EBRT), and trickling liquid velocity (TLV). The results revealed that the fast start-up time of about 10 days for VOCs removal was achieved in a BTF by using microbial intensive culture and synchronous acclimation. Moreover, under steady-state conditions, the removal efficiency (RE) of VOCs reached above 90.0% at the inlet VOC concentration of 150 mg/m3, EBRT of 45 s, and TLV of 1.5 m/h, corresponding to the elimination capacity (EC) of around 11.0 g m-3 h-1. Furthermore, both RSM-BBD and GA-BPANN models have the good ability to accurately simulate and predict the RE of VOCs under different experimental conditions. In addition, the kinetic behavior of VOCs biodegradation in BTFs conformed to zero-order model with diffusion limitation. The findings herein provide an efficient route for optimizing the parameters design of BFT, which facilitates the application of BTF in VOCs control for electronic manufactory industries.