Catalyst-driven predictive functions for Fischer–Tropsch selectivity and conversion

Abstract

The Fischer–Tropsch (FT) process remains a central pathway for producing low-carbon synthetic fuels from CO and CO2 feeds, yet predicting product distributions and conversion remains challenging because operating conditions and catalyst composition interact strongly. Existing empirical correlations often rely mainly on temperature, pressure, and feed ratio, which can limit their transferability across catalyst families and reaction conditions. This study develops catalyst-explicit response-surface models for CO-FT and CO2-FT using 176 literature-derived experiments spanning broad ranges of temperature, pressure, feed ratio, and binary catalyst-component descriptors. Closed-form predictive functions were obtained for CH4, C2–C4, and C5+ selectivities and CO/CO2 conversion. The models achieved R2 values up to 0.99 and Q2 values up to 0.98, while representing nonlinearities and catalyst–condition interactions. Coefficient analysis was interpreted in terms of statistical associations rather than direct mechanistic attribution, showing that reported catalyst-component presence and catalyst-condition interactions can shift predicted conversion and product selectivity in a product-window-dependent manner. Evaluation using an external literature dataset indicated that operating-variable-only correlations could not distinguish cases with identical operating conditions but different catalyst formulations, whereas the catalyst-explicit models allowed predictions to vary with the reported catalyst descriptors within the evaluated domain. Overall, the resulting functions provide a transparent catalyst-aware screening and process-modeling tool for evaluating FT selectivity and conversion trends within the calibrated literature domain.