Work

Data-driven analysis in the selective oligomerization of long-chain linear alpha olefin on zeolite catalysts: A machine learning-based parameter study

Year: 2024

Type: article

Abstract: In this study, the oligomerization of 1-octene using various zeolites was investigated through both experimental and machine learning (ML) methodologies. Experimental data were collected by varying the reaction temperature and using zeolites with different physicochemical properties. Conventional experimental approaches, including BET and TPD characterizations combined with GC analysis, were used to elucidate the effects of temperature, surface area, acidity, pore size, cavity size, and acid site location on the 1-octene oligomerization reaction. It is found that isomerization and oligomerization of 1-octene were in a competitive relationship, with the effectiveness of each parameter differing based on each reaction. For isomerization, structural properties such as pore size and cavity had a significant impact, leading to different preferential behaviors between kinetics and thermodynamics depending on the pore structure of the zeolites. In contrast, for the oligomerization reaction, surface area, temperature, and acidity were the dominant parameters. However, conventional experimental approaches exhibited clear limitations, as quantitative analysis of the reaction chemistry was not feasible. Therefore, SHAP and SISSO algorithms were employed to quantitatively elucidate the effectiveness of each reaction parameter on the oligomerization reaction. By utilizing the widely adopted SHAP method, the overall reaction chemistry and the importance of each parameter could be understandable. However, due to the limited amount of data, the variation in feature importance by temperature was not available. Therefore, the SISSO method, which only requires fewer data points and uncovers the black-box computation process, was utilized. With this approach, parameter importance changes at various reaction temperatures could be successfully observed.

Source: Fuel Processing Technology

DOI: 10.1016/j.fuproc.2024.108164

Authors: Sung Woo Lee, Marcel Jonathan Hidajat, Seung Hyeok Cha, Gwang-Nam Yun, Dong Won Hwang

Institutions: Korea Research Institute of Chemical Technology, Korea University of Science and Technology, Sungkyunkwan University

Language: English

Topic: Zeolite Catalysis and Synthesis

Subfield: Inorganic Chemistry

Field: Chemistry

Domain: Physical Sciences

Open Access status: gold

Estimated APC paid: $3,000

Funders: Korea Research Institute of Chemical Technology, National Research Foundation of Korea