The main goal of this study is to create a statistical and intelligent model that can forecast the removal efficiency of hydrogen sulfide (H2S) in a fixed bed biofilter. The majority of conventional models employ mathematical principles that necessitate highly technical infrastructure, extremely detailed information on the biofiltration mechanism, and estimation of model input parameters. However, in this study, Multiple Linear Regression and Support Vector Machine methodology were used to reduce the reliance on experimental data. Experimental research was done to determine the effects of three factors on H2S removal efficiency, including the moisture content of the compost used as the biofilter bed, the empty bed residence time in the reactor, and the H2S concentration in the biogas stream input to the biofilter. Experimental findings demonstrated that the biofilter with a moisture content of 6%, and empty bed residence time of 60 s was capable of removing up to 75% of H2S from a biogas stream with H2S inlet concentration of 70 ppm(v). With a Coefficient of Determination between 0.98 and 0.99, the proposed models show excellent ability to predict the H2S removal efficiency. Compared to a Multiple Linear Regression based model, test results from a Support Vector Machine based model showed better agreement with experimental data.

Keywords