Artificial neural network modeling of methane to C2s conversion process
Volume Title: 1
1Department of Chemical and Polymer Engineering, Islamic Azad University, South Tehran Branch, Tehran, Iran
2Department of Chemical and Polymer Engineering, Islamic Azad University, South Tehran Branch
The methane to ethylene oxidation is one of the promising methods for the conversion of natural gas to liquid fuels (GTL). In the oxidative coupling of methane (OCM) process, methane partially oxides to C2 hydrocarbons, COx, and water in the presence of an appropriate catalyst. Due to high COx production and low C2 hydrocarbons selectivity, different catalysts and operating conditions have been tested to achieve higher C2 yield. In this study, the OCM process was modeled utilizing an artificial neural network (ANN) using more than 600 experimental data. The reactor and catalyst type, temperature and CH4/O2 were assumed as input parameters hence the CH4 conversion was set as a target. The modeling was done by the mlp method and two functions of trainbr and trainlm. The comparison of two training functions was analyzed in terms of the difference between methane conversion percentage from experimental data and model predicted one. The error calculated from the tainbr and trainlm functions was 11% and 17% respectively, which revealed that trainbr would be a more reliable function for this study.