The Application of Hybrid RSM/ANN Methodology of an Iron-based Catalyst Performance in Fischer-Tropsch Synthesis

In this research, the performance and kinetics of an iron/manganese oxide catalyst in a fixed-bed reactor is studied through the fischer- tropsch synthesis. the ranges of the operating conditions are p: 1-12 barg, t: 513-553 k, h2/co ratio: 1-2 and ghsv: 4200-7000 (cm^3 (stp) / h /g cat) . the effects of the independent variables on the fischer-tropsch synthesis product are analyzed using a statistical model based on the experimental data. the response surface methodology and artificial neural network are used to model the experimental data for the carbon monoxide conversion (co% conversion) and co consumption rate. some of the statistical parameters such as correlation coefficient and mean square error are calculated to demonstrate the capability and sensitivity analysis of methods. the obtained results from the statistical methods show a good agreement with the experimental data. it is shown that the results of artificial neural network are more accurate than those of the response surface methodology. moreover, the optimal conditions are obtained from the maximum amount of co% conversion at p = 8 barg, t = 559.5 k, h2/co = 2.5 and ghsv = 7325.2 (cm^3 (stp) / h /g cat) while the maximum amount of consumption rate of co happens at p = 8 bar, t = 568 k, h2/co = 2.5 and ghsv = 2800 (cm^3 (stp) / h /g cat) . finally, the entire quadratic equations for all of the variables and optimal conditions for the responses are explained, respectively.