Tracking the green coke production when co-processing lipids at a commercial fluid catalytic cracker (FCC): combining isotope ¹⁴C and causal discovery analysis

Abstract

Co-processing biogenic feedstocks allows oil refiners to use their infrastructure while reducing the carbon intensity of the fuels they produce. Although policies such as British Columbia and California’s low carbon...

Interpretable Machine Learning with Physics-Based Feature Engineering: Application to the Catalytic Cracking Reaction of Polycyclic Aromatic Hydrocarbons

“…Recently, causal discovery methods, such as the linear non-Gaussian acyclic model (LiNGAM), have been proposed for linear models. 52,53 The statistical causality between descriptors selected by the LASSO model and product compositions will be explored in our future research. In addition, since the data set used in this study was not obtained with statistical analysis in mind, bias in the data is an issue.…”

“…Although we confirmed that descriptors associated with the reactions that have an important role in determining product composition are likely to be selected, the possibility of a pseudocorrelation has not been denied. Recently, causal discovery methods, such as the linear non-Gaussian acyclic model (LiNGAM), have been proposed for linear models. , The statistical causality between descriptors selected by the LASSO model and product compositions will be explored in our future research. In addition, since the data set used in this study was not obtained with statistical analysis in mind, bias in the data is an issue.…”