Feature-Based Machine Learning Modeling for Enzyme-Substrate Activity Prediction | AIChE

Feature-Based Machine Learning Modeling for Enzyme-Substrate Activity Prediction

Authors 

Liu, Z., Key Lab of Industrial Biocatalysis, Ministry of Education, Tsinghua University
Tar in biomass gasification synthetic gas is one of the factors that have the greatest influence on the stable operation of biomass gasification system and the utilization of synthetic gas. In this study, the catalytic steam reforming was evaluated to convert tar into hydrogen. Toluene, which accounts for a large content of biomass tar, was used as a model biomass tar. One of the biggest problems in catalytic steam reforming of hydrocarbon materials is the deactivation of catalyst due to carbon deposition. To overcome this, carbon support with high coking resistance was applied. A catalyst was prepared using activated carbon, carbon black (CB), coal, and nanocellulose as a support, Ni as an active metal, and Mg, K, and Ce as a promoter. Under constant conditions with a toluene concentration of 30 g/Nm3, the catalyst type, reaction temperature, steam/carbon molar ratio, and gas hourly space velocity effects were studied. In general, CB support improved catalyst reactivity because of its high thermal stability, mesoporous structure, and specific surface area, and the addition of a promoter to the Ni/CB catalyst increases the reactivity at low temperatures. All catalysts used in this study achieved H2 yield and carbon conversion of more than 70% and 87% without a reduction of catalyst before the reaction. The NiK/CB catalyst was most effective with a 100% carbon conversion rate and hydrogen yield at 550°C. The NiMg/CB catalyst confirmed stable reactivity without deactivation by coking for 50 hours.

Acknowledgment

This work was supported by the ERC Center funded by the National Research Foundation of Korea (NRF-2022R1A5A1033719).