Evaluation of metamodels for prediction of species concentration and reactor outlet temperature of Sabatier reactor

The work discussed in the poster features the investigation of the performance of metamodels such as polynomial regression, Random Forest, Auto-Encoder, Gradient Boosting Regressor and Gaussian Processes predicting the outlet temperature and species concentrations of a methane synthesis reactor. The metamodels are trained based on a detailed physical 1D model of the synthesis reactor, and the performance of the metamodels is evaluated by comparison with experimental results from a laboratory test bench for a Ni/Al2O3 catalyst.