A combined CFD modeling with population balance equation to predict pressure drop in venturi scrubbers

Prediction of Pressure Drop in Venturi Scrubbers by Multi-Gene Genetic Programming and Adaptive Neuro-Fuzzy Inference System

Studying the pressure drop in venturi scrubbers had been the subject of many types of researches due to its importance for removing pollutants from polluted gas. In this study, two new approaches based on Multi-Gene Genetic Programming (MGGP) and Adaptive Neuro-Fuzzy Inference System (ANFIS) were used to predict the pressure drop in venturi scrubbers. The main parameters studied were the throat gas velocity of venturi scrubbers (Vgth), the liquid to gas flow rate ratio (L/G), and the axial distance of the venturi scrubbers (z) as the inputs to the network, while the pressure drop was as the output. One set of experimental data, which was gathered from five different venturi scrubbers including a circular and an adjustable prismatic venturi scrubber with a wetted wall irrigation, a rectangular venturi scrubber and two ejector venturi scrubbers with different throat diameters were applied for this study. The results of ANFIS and MGGP were compared with experimental data and those values from Artificial Neural Networks (ANNs) from our previous work. In this work, the coefficient of the determination (i. e. R2 value) was used to show the prediction ability of these new approaches. Results showed that MGGP and ANFIS can accurately predict the pressure drop in venturi scrubbers with R2 values of 0.9972 and 0.9734, respectively. The results also showed that MGGP has more precision than ANFIS and ANNs. Therefore, based on MGGP, two correlations were generated for two clusters of data. The comparison results between one of these correlations (i. e. correlation 1 with R2 value equal to 0.9937) and other models showed that our correlation has a very good precision and can predict the pressure drop in a more agreement with the experimental data.