Predicting Removal Efficiency of Formaldehyde from Synthetic Contaminated Air in Biotrickling Filter Using Artificial Neural Network Modeling

Abstract

Formaldehyde (FA) is considered a toxic and mutagenic compound that is suspected to be carcinogenic for humans. FA is widely emitted to the atmosphere by several chemical industries. Therefore, it is important to have an effective system to remove it from air. Although biotrickling filter (BTF) has been introduced as a suitable method to remove FA from air, the optimum conditions have not yet been fully investigated in a satisfactory way. Here, the authors want to find the optimum conditions for effective factors, including pH, retention time, operation time, bed length, and volumetric air flow rate (VAFR) on a BTF. In this study, BTF was applied for treatment of FA from synthetically contaminated air. In order to predict FA removal efficiency (RE), artificial neural network (ANN) was used for simulation of BTF and for analyzing empirical data. ANN assessed RE and predicted data with acceptable root-mean-square error (RMSE) and correlation coefficient (R2). Moreover, a sensitivity analysis (SA) was performed showing that pH and operation time are effective at changing the amount of FA elimination. The results of this study can be used to operate BTF in the optimum conditions for obtaining high RE.