Predicting FTS products through artificial neural network modelling

Fischer-Tropsch synthesis is essential for converting CO2 into hydrocarbons, creating sustainable fuels and olefins. However, challenges in production yield and reaction kinetics remain. This study introduces an artificial neural network (ANN) to predict FT synthesis products from specific inputs, including temperature, pressure, GHSV, H2/CO2 ratio, and catalyst composition (Fe weight and K as a promoter). The ANN's ability to predict outputs like CH4, C2-4, C5+, CO2 conversion, and CO selectivity, without detailed reaction mechanisms, is a key innovation. This approach circumvents complex kinetic models. The network architecture is optimized for minimal error, and results are validated against a comprehensive database.